Skip to content

Instantly share code, notes, and snippets.

@boraseoksoon

boraseoksoon/ollama_models.json

Last active February 19, 2026 06:15
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save boraseoksoon/b8c5e2a44ec9cb01cbaa010a8953304c to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save boraseoksoon/b8c5e2a44ec9cb01cbaa010a8953304c to your computer and use it in GitHub Desktop.
Download ZIP
HLVM Ollama Model Catalog (auto-updated)
Raw
ollama_models.json
{
"models": [
{
"description": "Llama 3.1 is a new state-of-the-art model from Meta available in 8B, 70B and 405B parameter sizes.",
"id": "llama3.1",
"name": "Llama3.1",
"variants": [
{
"id": "llama3.1:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "llama3.1:405b",
"parameters": "405B",
"size": "243GB"
},
{
"id": "llama3.1:70b",
"parameters": "70B",
"size": "43GB"
},
{
"id": "llama3.1:8b",
"parameters": "8B",
"size": "4.9GB"
}
],
"vision": false,
"tools": true
},
{
"description": "DeepSeek-R1 is a family of open reasoning models with performance approaching that of leading models, such as O3 and Gemini 2.5 Pro.",
"id": "deepseek-r1",
"name": "Deepseek R1",
"variants": [
{
"id": "deepseek-r1:latest",
"parameters": "Unknown",
"size": "5.2GB"
},
{
"id": "deepseek-r1:671b",
"parameters": "671B",
"size": "404GB"
},
{
"id": "deepseek-r1:70b",
"parameters": "70B",
"size": "43GB"
},
{
"id": "deepseek-r1:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "deepseek-r1:14b",
"parameters": "14B",
"size": "9.0GB"
},
{
"id": "deepseek-r1:8b",
"parameters": "8B",
"size": "5.2GB"
},
{
"id": "deepseek-r1:7b",
"parameters": "7B",
"size": "4.7GB"
},
{
"id": "deepseek-r1:1.5b",
"parameters": "1.5B",
"size": "1.1GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "Meta's Llama 3.2 goes small with 1B and 3B models. ",
"id": "llama3.2",
"name": "Llama3.2",
"variants": [
{
"id": "llama3.2:latest",
"parameters": "Unknown",
"size": "2.0GB"
},
{
"id": "llama3.2:3b",
"parameters": "3B",
"size": "2.0GB"
},
{
"id": "llama3.2:1b",
"parameters": "1B",
"size": "1.3GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A high-performing open embedding model with a large token context window.",
"id": "nomic-embed-text",
"name": "Nomic Embed Text",
"variants": [
{
"id": "nomic-embed-text:latest",
"parameters": "Unknown",
"size": "274MB"
},
{
"id": "nomic-embed-text:137m-v1.5-fp16",
"parameters": "137M",
"size": "274MB"
},
{
"id": "nomic-embed-text:v1.5",
"parameters": "Unknown",
"size": "274MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "The current, most capable model that runs on a single GPU.",
"id": "gemma3",
"name": "Gemma3",
"variants": [
{
"id": "gemma3:latest",
"parameters": "Unknown",
"size": "3.3GB"
},
{
"id": "gemma3:270m",
"parameters": "270M",
"size": "292MB"
},
{
"id": "gemma3:27b",
"parameters": "27B",
"size": "17GB"
},
{
"id": "gemma3:27b-cloud",
"parameters": "27B",
"size": "Cloud (API only)"
},
{
"id": "gemma3:12b",
"parameters": "12B",
"size": "8.1GB"
},
{
"id": "gemma3:12b-cloud",
"parameters": "12B",
"size": "Cloud (API only)"
},
{
"id": "gemma3:4b",
"parameters": "4B",
"size": "3.3GB"
},
{
"id": "gemma3:4b-cloud",
"parameters": "4B",
"size": "Cloud (API only)"
},
{
"id": "gemma3:1b",
"parameters": "1B",
"size": "815MB"
}
],
"vision": true,
"cloud": true
},
{
"description": "The 7B model released by Mistral AI, updated to version 0.3.",
"id": "mistral",
"name": "Mistral",
"variants": [
{
"id": "mistral:latest",
"parameters": "Unknown",
"size": "4.4GB"
},
{
"id": "mistral:7b",
"parameters": "7B",
"size": "4.4GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Qwen2.5 models are pretrained on Alibaba's latest large-scale dataset, encompassing up to 18 trillion tokens. The model supports up to 128K tokens and has multilingual support. ",
"id": "qwen2.5",
"name": "Qwen2.5",
"variants": [
{
"id": "qwen2.5:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "qwen2.5:72b",
"parameters": "72B",
"size": "47GB"
},
{
"id": "qwen2.5:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "qwen2.5:14b",
"parameters": "14B",
"size": "9.0GB"
},
{
"id": "qwen2.5:7b",
"parameters": "7B",
"size": "4.7GB"
},
{
"id": "qwen2.5:3b",
"parameters": "3B",
"size": "1.9GB"
},
{
"id": "qwen2.5:1.5b",
"parameters": "1.5B",
"size": "986MB"
},
{
"id": "qwen2.5:0.5b",
"parameters": "0.5B",
"size": "398MB"
}
],
"vision": false,
"tools": true
},
{
"description": "Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.",
"id": "qwen3",
"name": "Qwen3",
"variants": [
{
"id": "qwen3:latest",
"parameters": "Unknown",
"size": "5.2GB"
},
{
"id": "qwen3:235b",
"parameters": "235B",
"size": "142GB"
},
{
"id": "qwen3:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "qwen3:30b",
"parameters": "30B",
"size": "19GB"
},
{
"id": "qwen3:14b",
"parameters": "14B",
"size": "9.3GB"
},
{
"id": "qwen3:8b",
"parameters": "8B",
"size": "5.2GB"
},
{
"id": "qwen3:4b",
"parameters": "4B",
"size": "2.5GB"
},
{
"id": "qwen3:1.7b",
"parameters": "1.7B",
"size": "1.4GB"
},
{
"id": "qwen3:0.6b",
"parameters": "0.6B",
"size": "523MB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "Google Gemma 2 is a high-performing and efficient model available in three sizes: 2B, 9B, and 27B.",
"id": "gemma2",
"name": "Gemma2",
"variants": [
{
"id": "gemma2:latest",
"parameters": "Unknown",
"size": "5.4GB"
},
{
"id": "gemma2:27b",
"parameters": "27B",
"size": "16GB"
},
{
"id": "gemma2:9b",
"parameters": "9B",
"size": "5.4GB"
},
{
"id": "gemma2:2b",
"parameters": "2B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "Phi-3 is a family of lightweight 3B (Mini) and 14B (Medium) state-of-the-art open models by Microsoft.",
"id": "phi3",
"name": "Phi3",
"variants": [
{
"id": "phi3:latest",
"parameters": "Unknown",
"size": "2.2GB"
},
{
"id": "phi3:14b",
"parameters": "14B",
"size": "7.9GB"
},
{
"id": "phi3:3.8b",
"parameters": "3.8B",
"size": "2.2GB"
}
],
"vision": false
},
{
"description": "Meta Llama 3: The most capable openly available LLM to date",
"id": "llama3",
"name": "Llama3",
"variants": [
{
"id": "llama3:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama3:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "llama3:8b",
"parameters": "8B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "🌋 LLaVA is a novel end-to-end trained large multimodal model that combines a vision encoder and Vicuna for general-purpose visual and language understanding. Updated to version 1.6.",
"id": "llava",
"name": "Llava",
"variants": [
{
"id": "llava:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llava:34b",
"parameters": "34B",
"size": "20GB"
},
{
"id": "llava:13b",
"parameters": "13B",
"size": "8.0GB"
},
{
"id": "llava:7b",
"parameters": "7B",
"size": "4.7GB"
}
],
"vision": true
},
{
"description": "The latest series of Code-Specific Qwen models, with significant improvements in code generation, code reasoning, and code fixing.",
"id": "qwen2.5-coder",
"name": "Qwen2.5 Coder",
"variants": [
{
"id": "qwen2.5-coder:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "qwen2.5-coder:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "qwen2.5-coder:14b",
"parameters": "14B",
"size": "9.0GB"
},
{
"id": "qwen2.5-coder:7b",
"parameters": "7B",
"size": "4.7GB"
},
{
"id": "qwen2.5-coder:3b",
"parameters": "3B",
"size": "1.9GB"
},
{
"id": "qwen2.5-coder:1.5b",
"parameters": "1.5B",
"size": "986MB"
},
{
"id": "qwen2.5-coder:0.5b",
"parameters": "0.5B",
"size": "398MB"
}
],
"vision": false,
"tools": true
},
{
"description": "State-of-the-art large embedding model from mixedbread.ai",
"id": "mxbai-embed-large",
"name": "Mxbai Embed Large",
"variants": [
{
"id": "mxbai-embed-large:latest",
"parameters": "Unknown",
"size": "670MB"
},
{
"id": "mxbai-embed-large:335m",
"parameters": "335M",
"size": "670MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "Phi-4 is a 14B parameter, state-of-the-art open model from Microsoft.",
"id": "phi4",
"name": "Phi4",
"variants": [
{
"id": "phi4:latest",
"parameters": "Unknown",
"size": "9.1GB"
},
{
"id": "phi4:14b",
"parameters": "14B",
"size": "9.1GB"
}
],
"vision": false
},
{
"description": "OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases.",
"id": "gpt-oss",
"name": "Gpt Oss",
"variants": [
{
"id": "gpt-oss:latest",
"parameters": "Unknown",
"size": "14GB"
},
{
"id": "gpt-oss:120b",
"parameters": "120B",
"size": "65GB"
},
{
"id": "gpt-oss:120b-cloud",
"parameters": "120B",
"size": "Cloud (API only)"
},
{
"id": "gpt-oss:20b",
"parameters": "20B",
"size": "14GB"
},
{
"id": "gpt-oss:20b-cloud",
"parameters": "20B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Gemma is a family of lightweight, state-of-the-art open models built by Google DeepMind. Updated to version 1.1",
"id": "gemma",
"name": "Gemma",
"variants": [
{
"id": "gemma:latest",
"parameters": "Unknown",
"size": "5.0GB"
},
{
"id": "gemma:7b",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "gemma:2b",
"parameters": "2B",
"size": "1.7GB"
}
],
"vision": false
},
{
"description": "Qwen 1.5 is a series of large language models by Alibaba Cloud spanning from 0.5B to 110B parameters",
"id": "qwen",
"name": "Qwen",
"variants": [
{
"id": "qwen:latest",
"parameters": "Unknown",
"size": "2.3GB"
},
{
"id": "qwen:110b",
"parameters": "110B",
"size": "63GB"
},
{
"id": "qwen:72b",
"parameters": "72B",
"size": "41GB"
},
{
"id": "qwen:32b",
"parameters": "32B",
"size": "18GB"
},
{
"id": "qwen:14b",
"parameters": "14B",
"size": "8.2GB"
},
{
"id": "qwen:7b",
"parameters": "7B",
"size": "4.5GB"
},
{
"id": "qwen:4b",
"parameters": "4B",
"size": "2.3GB"
},
{
"id": "qwen:1.8b",
"parameters": "1.8B",
"size": "1.1GB"
},
{
"id": "qwen:0.5b",
"parameters": "0.5B",
"size": "395MB"
}
],
"vision": false
},
{
"description": "Llama 2 is a collection of foundation language models ranging from 7B to 70B parameters.",
"id": "llama2",
"name": "Llama2",
"variants": [
{
"id": "llama2:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "llama2:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "llama2:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "llama2:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Qwen2 is a new series of large language models from Alibaba group",
"id": "qwen2",
"name": "Qwen2",
"variants": [
{
"id": "qwen2:latest",
"parameters": "Unknown",
"size": "4.4GB"
},
{
"id": "qwen2:72b",
"parameters": "72B",
"size": "41GB"
},
{
"id": "qwen2:7b",
"parameters": "7B",
"size": "4.4GB"
},
{
"id": "qwen2:1.5b",
"parameters": "1.5B",
"size": "935MB"
},
{
"id": "qwen2:0.5b",
"parameters": "0.5B",
"size": "352MB"
}
],
"vision": false,
"tools": true
},
{
"description": "A series of multimodal LLMs (MLLMs) designed for vision-language understanding.",
"id": "minicpm-v",
"name": "Minicpm V",
"variants": [
{
"id": "minicpm-v:latest",
"parameters": "Unknown",
"size": "5.5GB"
},
{
"id": "minicpm-v:8b",
"parameters": "8B",
"size": "5.5GB"
}
],
"vision": true
},
{
"description": "A large language model that can use text prompts to generate and discuss code.",
"id": "codellama",
"name": "Codellama",
"variants": [
{
"id": "codellama:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "codellama:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "codellama:34b",
"parameters": "34B",
"size": "19GB"
},
{
"id": "codellama:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "codellama:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Llama 3.2 Vision is a collection of instruction-tuned image reasoning generative models in 11B and 90B sizes.",
"id": "llama3.2-vision",
"name": "Llama3.2 Vision",
"variants": [
{
"id": "llama3.2-vision:latest",
"parameters": "Unknown",
"size": "7.8GB"
},
{
"id": "llama3.2-vision:90b",
"parameters": "90B",
"size": "55GB"
},
{
"id": "llama3.2-vision:11b",
"parameters": "11B",
"size": "7.8GB"
}
],
"vision": true
},
{
"description": "The TinyLlama project is an open endeavor to train a compact 1.1B Llama model on 3 trillion tokens.",
"id": "tinyllama",
"name": "Tinyllama",
"variants": [
{
"id": "tinyllama:latest",
"parameters": "Unknown",
"size": "638MB"
},
{
"id": "tinyllama:1.1b",
"parameters": "1.1B",
"size": "638MB"
}
],
"vision": false
},
{
"description": "Dolphin 3.0 Llama 3.1 8B 🐬 is the next generation of the Dolphin series of instruct-tuned models designed to be the ultimate general purpose local model, enabling coding, math, agentic, function calling, and general use cases.",
"id": "dolphin3",
"name": "Dolphin3",
"variants": [
{
"id": "dolphin3:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "dolphin3:8b",
"parameters": "8B",
"size": "4.9GB"
}
],
"vision": false
},
{
"description": "A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.",
"id": "deepseek-v3",
"name": "Deepseek V3",
"variants": [
{
"id": "deepseek-v3:latest",
"parameters": "Unknown",
"size": "404GB"
},
{
"id": "deepseek-v3:671b",
"parameters": "671B",
"size": "404GB"
}
],
"vision": false
},
{
"description": "OLMo 2 is a new family of 7B and 13B models trained on up to 5T tokens. These models are on par with or better than equivalently sized fully open models, and competitive with open-weight models such as Llama 3.1 on English academic benchmarks.",
"id": "olmo2",
"name": "Olmo2",
"variants": [
{
"id": "olmo2:latest",
"parameters": "Unknown",
"size": "4.5GB"
},
{
"id": "olmo2:13b",
"parameters": "13B",
"size": "8.4GB"
},
{
"id": "olmo2:7b",
"parameters": "7B",
"size": "4.5GB"
}
],
"vision": false
},
{
"description": "A state-of-the-art 12B model with 128k context length, built by Mistral AI in collaboration with NVIDIA.",
"id": "mistral-nemo",
"name": "Mistral Nemo",
"variants": [
{
"id": "mistral-nemo:latest",
"parameters": "Unknown",
"size": "7.1GB"
},
{
"id": "mistral-nemo:12b",
"parameters": "12B",
"size": "7.1GB"
}
],
"vision": false,
"tools": true
},
{
"description": "BGE-M3 is a new model from BAAI distinguished for its versatility in Multi-Functionality, Multi-Linguality, and Multi-Granularity.",
"id": "bge-m3",
"name": "Bge M3",
"variants": [
{
"id": "bge-m3:latest",
"parameters": "Unknown",
"size": "1.2GB"
},
{
"id": "bge-m3:567m",
"parameters": "567M",
"size": "1.2GB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "New state of the art 70B model. Llama 3.3 70B offers similar performance compared to the Llama 3.1 405B model.",
"id": "llama3.3",
"name": "Llama3.3",
"variants": [
{
"id": "llama3.3:latest",
"parameters": "Unknown",
"size": "43GB"
},
{
"id": "llama3.3:70b",
"parameters": "70B",
"size": "43GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Alibaba's performant long context models for agentic and coding tasks.",
"id": "qwen3-coder",
"name": "Qwen3 Coder",
"variants": [
{
"id": "qwen3-coder:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "qwen3-coder:480b",
"parameters": "480B",
"size": "290GB"
},
{
"id": "qwen3-coder:480b-cloud",
"parameters": "480B",
"size": "Cloud (API only)"
},
{
"id": "qwen3-coder:30b",
"parameters": "30B",
"size": "19GB"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "DeepSeek Coder is a capable coding model trained on two trillion code and natural language tokens.",
"id": "deepseek-coder",
"name": "Deepseek Coder",
"variants": [
{
"id": "deepseek-coder:latest",
"parameters": "Unknown",
"size": "776MB"
},
{
"id": "deepseek-coder:33b",
"parameters": "33B",
"size": "19GB"
},
{
"id": "deepseek-coder:6.7b",
"parameters": "6.7B",
"size": "3.8GB"
},
{
"id": "deepseek-coder:1.3b",
"parameters": "1.3B",
"size": "776MB"
}
],
"vision": false
},
{
"description": "SmolLM2 is a family of compact language models available in three size: 135M, 360M, and 1.7B parameters.",
"id": "smollm2",
"name": "Smollm2",
"variants": [
{
"id": "smollm2:latest",
"parameters": "Unknown",
"size": "1.8GB"
},
{
"id": "smollm2:360m",
"parameters": "360M",
"size": "726MB"
},
{
"id": "smollm2:135m",
"parameters": "135M",
"size": "271MB"
},
{
"id": "smollm2:1.7b",
"parameters": "1.7B",
"size": "1.8GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Embedding models on very large sentence level datasets.",
"id": "all-minilm",
"name": "All Minilm",
"variants": [
{
"id": "all-minilm:latest",
"parameters": "Unknown",
"size": "46MB"
},
{
"id": "all-minilm:33m",
"parameters": "33M",
"size": "67MB"
},
{
"id": "all-minilm:22m",
"parameters": "22M",
"size": "46MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "Mistral Small 3 sets a new benchmark in the “small” Large Language Models category below 70B.",
"id": "mistral-small",
"name": "Mistral Small",
"variants": [
{
"id": "mistral-small:latest",
"parameters": "Unknown",
"size": "14GB"
},
{
"id": "mistral-small:24b",
"parameters": "24B",
"size": "14GB"
},
{
"id": "mistral-small:22b",
"parameters": "22B",
"size": "13GB"
}
],
"vision": false,
"tools": true
},
{
"description": "CodeGemma is a collection of powerful, lightweight models that can perform a variety of coding tasks like fill-in-the-middle code completion, code generation, natural language understanding, mathematical reasoning, and instruction following.",
"id": "codegemma",
"name": "Codegemma",
"variants": [
{
"id": "codegemma:latest",
"parameters": "Unknown",
"size": "5.0GB"
},
{
"id": "codegemma:7b",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "codegemma:2b",
"parameters": "2B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "A family of efficient AI models under 10B parameters performant in science, math, and coding through innovative training techniques.",
"id": "falcon3",
"name": "Falcon3",
"variants": [
{
"id": "falcon3:latest",
"parameters": "Unknown",
"size": "4.6GB"
},
{
"id": "falcon3:10b",
"parameters": "10B",
"size": "6.3GB"
},
{
"id": "falcon3:7b",
"parameters": "7B",
"size": "4.6GB"
},
{
"id": "falcon3:3b",
"parameters": "3B",
"size": "2.0GB"
},
{
"id": "falcon3:1b",
"parameters": "1B",
"size": "1.8GB"
}
],
"vision": false
},
{
"description": "The IBM Granite 1B and 3B models are long-context mixture of experts (MoE) Granite models from IBM designed for low latency usage.",
"id": "granite3.1-moe",
"name": "Granite3.1 Moe",
"variants": [
{
"id": "granite3.1-moe:latest",
"parameters": "Unknown",
"size": "2.0GB"
},
{
"id": "granite3.1-moe:3b",
"parameters": "3B",
"size": "2.0GB"
},
{
"id": "granite3.1-moe:1b",
"parameters": "1B",
"size": "1.4GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A LLaVA model fine-tuned from Llama 3 Instruct with better scores in several benchmarks.",
"id": "llava-llama3",
"name": "Llava Llama3",
"variants": [
{
"id": "llava-llama3:latest",
"parameters": "Unknown",
"size": "5.5GB"
},
{
"id": "llava-llama3:8b",
"parameters": "8B",
"size": "5.5GB"
}
],
"vision": true
},
{
"description": "StarCoder2 is the next generation of transparently trained open code LLMs that comes in three sizes: 3B, 7B and 15B parameters. ",
"id": "starcoder2",
"name": "Starcoder2",
"variants": [
{
"id": "starcoder2:latest",
"parameters": "Unknown",
"size": "1.7GB"
},
{
"id": "starcoder2:15b",
"parameters": "15B",
"size": "9.1GB"
},
{
"id": "starcoder2:7b",
"parameters": "7B",
"size": "4.0GB"
},
{
"id": "starcoder2:3b",
"parameters": "3B",
"size": "1.7GB"
}
],
"vision": false
},
{
"description": "A general-purpose model ranging from 3 billion parameters to 70 billion, suitable for entry-level hardware.",
"id": "orca-mini",
"name": "Orca Mini",
"variants": [
{
"id": "orca-mini:latest",
"parameters": "Unknown",
"size": "2.0GB"
},
{
"id": "orca-mini:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "orca-mini:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "orca-mini:7b",
"parameters": "7B",
"size": "3.8GB"
},
{
"id": "orca-mini:3b",
"parameters": "3B",
"size": "2.0GB"
}
],
"vision": false
},
{
"description": "QwQ is the reasoning model of the Qwen series.",
"id": "qwq",
"name": "Qwq",
"variants": [
{
"id": "qwq:latest",
"parameters": "Unknown",
"size": "20GB"
},
{
"id": "qwq:32b",
"parameters": "32B",
"size": "20GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A suite of text embedding models by Snowflake, optimized for performance.",
"id": "snowflake-arctic-embed",
"name": "Snowflake Arctic Embed",
"variants": [
{
"id": "snowflake-arctic-embed:latest",
"parameters": "Unknown",
"size": "669MB"
},
{
"id": "snowflake-arctic-embed:335m",
"parameters": "335M",
"size": "669MB"
},
{
"id": "snowflake-arctic-embed:137m",
"parameters": "137M",
"size": "274MB"
},
{
"id": "snowflake-arctic-embed:110m",
"parameters": "110M",
"size": "219MB"
},
{
"id": "snowflake-arctic-embed:33m",
"parameters": "33M",
"size": "67MB"
},
{
"id": "snowflake-arctic-embed:22m",
"parameters": "22M",
"size": "46MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "A set of Mixture of Experts (MoE) model with open weights by Mistral AI in 8x7b and 8x22b parameter sizes.",
"id": "mixtral",
"name": "Mixtral",
"variants": [
{
"id": "mixtral:latest",
"parameters": "Unknown",
"size": "26GB"
},
{
"id": "mixtral:8x22b",
"parameters": "22B",
"size": "80GB"
},
{
"id": "mixtral:8x7b",
"parameters": "7B",
"size": "26GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Uncensored Llama 2 model by George Sung and Jarrad Hope.",
"id": "llama2-uncensored",
"name": "Llama2 Uncensored",
"variants": [
{
"id": "llama2-uncensored:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "llama2-uncensored:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "llama2-uncensored:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "An open-source Mixture-of-Experts code language model that achieves performance comparable to GPT4-Turbo in code-specific tasks.",
"id": "deepseek-coder-v2",
"name": "Deepseek Coder V2",
"variants": [
{
"id": "deepseek-coder-v2:latest",
"parameters": "Unknown",
"size": "8.9GB"
},
{
"id": "deepseek-coder-v2:236b",
"parameters": "236B",
"size": "133GB"
},
{
"id": "deepseek-coder-v2:16b",
"parameters": "16B",
"size": "8.9GB"
}
],
"vision": false
},
{
"description": "The most powerful vision-language model in the Qwen model family to date. ",
"id": "qwen3-vl",
"name": "Qwen3 Vl",
"variants": [
{
"id": "qwen3-vl:latest",
"parameters": "Unknown",
"size": "6.1GB"
},
{
"id": "qwen3-vl:235b",
"parameters": "235B",
"size": "143GB"
},
{
"id": "qwen3-vl:235b-cloud",
"parameters": "235B",
"size": "Cloud (API only)"
},
{
"id": "qwen3-vl:235b-instruct-cloud",
"parameters": "235B",
"size": "Cloud (API only)"
},
{
"id": "qwen3-vl:32b",
"parameters": "32B",
"size": "21GB"
},
{
"id": "qwen3-vl:30b",
"parameters": "30B",
"size": "20GB"
},
{
"id": "qwen3-vl:8b",
"parameters": "8B",
"size": "6.1GB"
},
{
"id": "qwen3-vl:4b",
"parameters": "4B",
"size": "3.3GB"
},
{
"id": "qwen3-vl:2b",
"parameters": "2B",
"size": "1.9GB"
}
],
"vision": true,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Cogito v1 Preview is a family of hybrid reasoning models by Deep Cogito that outperform the best available open models of the same size, including counterparts from LLaMA, DeepSeek, and Qwen across most standard benchmarks.",
"id": "cogito",
"name": "Cogito",
"variants": [
{
"id": "cogito:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "cogito:70b",
"parameters": "70B",
"size": "43GB"
},
{
"id": "cogito:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "cogito:14b",
"parameters": "14B",
"size": "9.0GB"
},
{
"id": "cogito:8b",
"parameters": "8B",
"size": "4.9GB"
},
{
"id": "cogito:3b",
"parameters": "3B",
"size": "2.2GB"
}
],
"vision": false,
"tools": true
},
{
"description": "An update to Mistral Small that improves on function calling, instruction following, and less repetition errors.",
"id": "mistral-small3.2",
"name": "Mistral Small3.2",
"variants": [
{
"id": "mistral-small3.2:latest",
"parameters": "Unknown",
"size": "15GB"
},
{
"id": "mistral-small3.2:24b",
"parameters": "24B",
"size": "15GB"
}
],
"vision": true,
"tools": true
},
{
"description": "Flagship vision-language model of Qwen and also a significant leap from the previous Qwen2-VL.",
"id": "qwen2.5vl",
"name": "Qwen2.5vl",
"variants": [
{
"id": "qwen2.5vl:latest",
"parameters": "Unknown",
"size": "6.0GB"
},
{
"id": "qwen2.5vl:72b",
"parameters": "72B",
"size": "49GB"
},
{
"id": "qwen2.5vl:32b",
"parameters": "32B",
"size": "21GB"
},
{
"id": "qwen2.5vl:7b",
"parameters": "7B",
"size": "6.0GB"
},
{
"id": "qwen2.5vl:3b",
"parameters": "3B",
"size": "3.2GB"
}
],
"vision": true
},
{
"description": "2.7B uncensored Dolphin model by Eric Hartford, based on the Phi language model by Microsoft Research.",
"id": "dolphin-phi",
"name": "Dolphin Phi",
"variants": [
{
"id": "dolphin-phi:latest",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "dolphin-phi:2.7b",
"parameters": "2.7B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "Gemma 3n models are designed for efficient execution on everyday devices such as laptops, tablets or phones. ",
"id": "gemma3n",
"name": "Gemma3n",
"variants": [
{
"id": "gemma3n:latest",
"parameters": "Unknown",
"size": "7.5GB"
},
{
"id": "gemma3n:e4b",
"parameters": "4B",
"size": "7.5GB"
},
{
"id": "gemma3n:e2b",
"parameters": "2B",
"size": "5.6GB"
}
],
"vision": false
},
{
"description": "Meta's latest collection of multimodal models.",
"id": "llama4",
"name": "Llama4",
"variants": [
{
"id": "llama4:latest",
"parameters": "Unknown",
"size": "67GB"
},
{
"id": "llama4:16x17b",
"parameters": "17B",
"size": "67GB"
},
{
"id": "llama4:128x17b",
"parameters": "17B",
"size": "245GB"
}
],
"vision": true,
"tools": true
},
{
"description": "Phi 4 reasoning and reasoning plus are 14-billion parameter open-weight reasoning models that rival much larger models on complex reasoning tasks. ",
"id": "phi4-reasoning",
"name": "Phi4 Reasoning",
"variants": [
{
"id": "phi4-reasoning:latest",
"parameters": "Unknown",
"size": "11GB"
},
{
"id": "phi4-reasoning:14b",
"parameters": "14B",
"size": "11GB"
}
],
"vision": false
},
{
"description": "A fine-tuned version of Deepseek-R1-Distilled-Qwen-1.5B that surpasses the performance of OpenAI’s o1-preview with just 1.5B parameters on popular math evaluations.",
"id": "deepscaler",
"name": "Deepscaler",
"variants": [
{
"id": "deepscaler:latest",
"parameters": "Unknown",
"size": "3.6GB"
},
{
"id": "deepscaler:1.5b",
"parameters": "1.5B",
"size": "3.6GB"
}
],
"vision": false
},
{
"description": "Magistral is a small, efficient reasoning model with 24B parameters.",
"id": "magistral",
"name": "Magistral",
"variants": [
{
"id": "magistral:latest",
"parameters": "Unknown",
"size": "14GB"
},
{
"id": "magistral:24b",
"parameters": "24B",
"size": "14GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "Dolphin 2.9 is a new model with 8B and 70B sizes by Eric Hartford based on Llama 3 that has a variety of instruction, conversational, and coding skills.",
"id": "dolphin-llama3",
"name": "Dolphin Llama3",
"variants": [
{
"id": "dolphin-llama3:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "dolphin-llama3:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "dolphin-llama3:8b",
"parameters": "8B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "Uncensored, 8x7b and 8x22b fine-tuned models based on the Mixtral mixture of experts models that excels at coding tasks. Created by Eric Hartford.",
"id": "dolphin-mixtral",
"name": "Dolphin Mixtral",
"variants": [
{
"id": "dolphin-mixtral:latest",
"parameters": "Unknown",
"size": "26GB"
},
{
"id": "dolphin-mixtral:8x22b",
"parameters": "22B",
"size": "80GB"
},
{
"id": "dolphin-mixtral:8x7b",
"parameters": "7B",
"size": "26GB"
}
],
"vision": false
},
{
"description": "Phi-2: a 2.7B language model by Microsoft Research that demonstrates outstanding reasoning and language understanding capabilities.",
"id": "phi",
"name": "Phi",
"variants": [
{
"id": "phi:latest",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "phi:2.7b",
"parameters": "2.7B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "🪐 A family of small models with 135M, 360M, and 1.7B parameters, trained on a new high-quality dataset.",
"id": "smollm",
"name": "Smollm",
"variants": [
{
"id": "smollm:latest",
"parameters": "Unknown",
"size": "991MB"
},
{
"id": "smollm:360m",
"parameters": "360M",
"size": "229MB"
},
{
"id": "smollm:135m",
"parameters": "135M",
"size": "92MB"
},
{
"id": "smollm:1.7b",
"parameters": "1.7B",
"size": "991MB"
}
],
"vision": false
},
{
"description": "IBM Granite 2B and 8B models are 128K context length language models that have been fine-tuned for improved reasoning and instruction-following capabilities.",
"id": "granite3.3",
"name": "Granite3.3",
"variants": [
{
"id": "granite3.3:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "granite3.3:8b",
"parameters": "8B",
"size": "4.9GB"
},
{
"id": "granite3.3:2b",
"parameters": "2B",
"size": "1.5GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "Phi-4-mini brings significant enhancements in multilingual support, reasoning, and mathematics, and now, the long-awaited function calling feature is finally supported.",
"id": "phi4-mini",
"name": "Phi4 Mini",
"variants": [
{
"id": "phi4-mini:latest",
"parameters": "Unknown",
"size": "2.5GB"
},
{
"id": "phi4-mini:3.8b",
"parameters": "3.8B",
"size": "2.5GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Building upon the foundational models of the Qwen3 series, Qwen3 Embedding provides a comprehensive range of text embeddings models in various sizes",
"id": "qwen3-embedding",
"name": "Qwen3 Embedding",
"variants": [
{
"id": "qwen3-embedding:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "qwen3-embedding:8b",
"parameters": "8B",
"size": "4.7GB"
},
{
"id": "qwen3-embedding:4b",
"parameters": "4B",
"size": "2.5GB"
},
{
"id": "qwen3-embedding:0.6b",
"parameters": "0.6B",
"size": "639MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "Codestral is Mistral AI’s first-ever code model designed for code generation tasks.",
"id": "codestral",
"name": "Codestral",
"variants": [
{
"id": "codestral:latest",
"parameters": "Unknown",
"size": "13GB"
},
{
"id": "codestral:22b",
"parameters": "22B",
"size": "13GB"
}
],
"vision": false
},
{
"description": "A fully open-source family of reasoning models built using a dataset derived by distilling DeepSeek-R1.",
"id": "openthinker",
"name": "Openthinker",
"variants": [
{
"id": "openthinker:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "openthinker:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "openthinker:7b",
"parameters": "7B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "A compact and efficient vision-language model, specifically designed for visual document understanding, enabling automated content extraction from tables, charts, infographics, plots, diagrams, and more.",
"id": "granite3.2-vision",
"name": "Granite3.2 Vision",
"variants": [
{
"id": "granite3.2-vision:latest",
"parameters": "Unknown",
"size": "2.4GB"
},
{
"id": "granite3.2-vision:2b",
"parameters": "2B",
"size": "2.4GB"
}
],
"vision": true,
"tools": true
},
{
"description": "Devstral: the best open source model for coding agents",
"id": "devstral",
"name": "Devstral",
"variants": [
{
"id": "devstral:latest",
"parameters": "Unknown",
"size": "14GB"
},
{
"id": "devstral:24b",
"parameters": "24B",
"size": "14GB"
}
],
"vision": false,
"tools": true
},
{
"description": "The uncensored Dolphin model based on Mistral that excels at coding tasks. Updated to version 2.8.",
"id": "dolphin-mistral",
"name": "Dolphin Mistral",
"variants": [
{
"id": "dolphin-mistral:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "dolphin-mistral:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Granite 4 features improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.",
"id": "granite4",
"name": "Granite4",
"variants": [
{
"id": "granite4:latest",
"parameters": "Unknown",
"size": "2.1GB"
},
{
"id": "granite4:350m",
"parameters": "350M",
"size": "708MB"
},
{
"id": "granite4:3b",
"parameters": "3B",
"size": "2.1GB"
},
{
"id": "granite4:1b",
"parameters": "1B",
"size": "3.3GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Command R is a Large Language Model optimized for conversational interaction and long context tasks.",
"id": "command-r",
"name": "Command R",
"variants": [
{
"id": "command-r:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "command-r:35b",
"parameters": "35B",
"size": "19GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A family of open foundation models by IBM for Code Intelligence",
"id": "granite-code",
"name": "Granite Code",
"variants": [
{
"id": "granite-code:latest",
"parameters": "Unknown",
"size": "2.0GB"
},
{
"id": "granite-code:34b",
"parameters": "34B",
"size": "19GB"
},
{
"id": "granite-code:20b",
"parameters": "20B",
"size": "12GB"
},
{
"id": "granite-code:8b",
"parameters": "8B",
"size": "4.6GB"
},
{
"id": "granite-code:3b",
"parameters": "3B",
"size": "2.0GB"
}
],
"vision": false
},
{
"description": "State of the art large language model from Microsoft AI with improved performance on complex chat, multilingual, reasoning and agent use cases.",
"id": "wizardlm2",
"name": "Wizardlm2",
"variants": [
{
"id": "wizardlm2:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "wizardlm2:8x22b",
"parameters": "22B",
"size": "80GB"
},
{
"id": "wizardlm2:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "DeepCoder is a fully open-Source 14B coder model at O3-mini level, with a 1.5B version also available.",
"id": "deepcoder",
"name": "Deepcoder",
"variants": [
{
"id": "deepcoder:latest",
"parameters": "Unknown",
"size": "9.0GB"
},
{
"id": "deepcoder:14b",
"parameters": "14B",
"size": "9.0GB"
},
{
"id": "deepcoder:1.5b",
"parameters": "1.5B",
"size": "1.1GB"
}
],
"vision": false
},
{
"description": "moondream2 is a small vision language model designed to run efficiently on edge devices.",
"id": "moondream",
"name": "Moondream",
"variants": [
{
"id": "moondream:latest",
"parameters": "Unknown",
"size": "1.7GB"
},
{
"id": "moondream:1.8b",
"parameters": "1.8B",
"size": "1.7GB"
}
],
"vision": true
},
{
"description": "Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research",
"id": "hermes3",
"name": "Hermes3",
"variants": [
{
"id": "hermes3:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "hermes3:405b",
"parameters": "405B",
"size": "229GB"
},
{
"id": "hermes3:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "hermes3:8b",
"parameters": "8B",
"size": "4.7GB"
},
{
"id": "hermes3:3b",
"parameters": "3B",
"size": "2.0GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Building upon Mistral Small 3, Mistral Small 3.1 (2503) adds state-of-the-art vision understanding and enhances long context capabilities up to 128k tokens without compromising text performance.",
"id": "mistral-small3.1",
"name": "Mistral Small3.1",
"variants": [
{
"id": "mistral-small3.1:latest",
"parameters": "Unknown",
"size": "15GB"
},
{
"id": "mistral-small3.1:24b",
"parameters": "24B",
"size": "15GB"
}
],
"vision": true,
"tools": true
},
{
"description": "LFM2.5 is a new family of hybrid models designed for on-device deployment.",
"id": "lfm2.5-thinking",
"name": "Lfm2.5 Thinking",
"variants": [
{
"id": "lfm2.5-thinking:latest",
"parameters": "Unknown",
"size": "731MB"
},
{
"id": "lfm2.5-thinking:1.2b",
"parameters": "1.2B",
"size": "731MB"
}
],
"vision": false,
"tools": true
},
{
"description": "Yi 1.5 is a high-performing, bilingual language model.",
"id": "yi",
"name": "Yi",
"variants": [
{
"id": "yi:latest",
"parameters": "Unknown",
"size": "3.5GB"
},
{
"id": "yi:34b",
"parameters": "34B",
"size": "19GB"
},
{
"id": "yi:9b",
"parameters": "9B",
"size": "5.0GB"
},
{
"id": "yi:6b",
"parameters": "6B",
"size": "3.5GB"
}
],
"vision": false
},
{
"description": "Zephyr is a series of fine-tuned versions of the Mistral and Mixtral models that are trained to act as helpful assistants.",
"id": "zephyr",
"name": "Zephyr",
"variants": [
{
"id": "zephyr:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "zephyr:141b",
"parameters": "141B",
"size": "80GB"
},
{
"id": "zephyr:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Mistral Large 2 is Mistral's new flagship model that is significantly more capable in code generation, mathematics, and reasoning with 128k context window and support for dozens of languages.",
"id": "mistral-large",
"name": "Mistral Large",
"variants": [
{
"id": "mistral-large:latest",
"parameters": "Unknown",
"size": "73GB"
},
{
"id": "mistral-large:123b",
"parameters": "123B",
"size": "73GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A lightweight AI model with 3.8 billion parameters with performance overtaking similarly and larger sized models.",
"id": "phi3.5",
"name": "Phi3.5",
"variants": [
{
"id": "phi3.5:latest",
"parameters": "Unknown",
"size": "2.2GB"
},
{
"id": "phi3.5:3.8b",
"parameters": "3.8B",
"size": "2.2GB"
}
],
"vision": false
},
{
"description": "Wizard Vicuna Uncensored is a 7B, 13B, and 30B parameter model based on Llama 2 uncensored by Eric Hartford.",
"id": "wizard-vicuna-uncensored",
"name": "Wizard Vicuna Uncensored",
"variants": [
{
"id": "wizard-vicuna-uncensored:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "wizard-vicuna-uncensored:30b",
"parameters": "30B",
"size": "18GB"
},
{
"id": "wizard-vicuna-uncensored:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizard-vicuna-uncensored:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "EmbeddingGemma is a 300M parameter embedding model from Google.",
"id": "embeddinggemma",
"name": "Embeddinggemma",
"variants": [
{
"id": "embeddinggemma:latest",
"parameters": "Unknown",
"size": "622MB"
},
{
"id": "embeddinggemma:300m",
"parameters": "300M",
"size": "622MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "BakLLaVA is a multimodal model consisting of the Mistral 7B base model augmented with the LLaVA architecture.",
"id": "bakllava",
"name": "Bakllava",
"variants": [
{
"id": "bakllava:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "bakllava:7b",
"parameters": "7B",
"size": "4.7GB"
}
],
"vision": true
},
{
"description": "StarCoder is a code generation model trained on 80+ programming languages.",
"id": "starcoder",
"name": "Starcoder",
"variants": [
{
"id": "starcoder:latest",
"parameters": "Unknown",
"size": "1.8GB"
},
{
"id": "starcoder:15b",
"parameters": "15B",
"size": "9.0GB"
},
{
"id": "starcoder:7b",
"parameters": "7B",
"size": "4.3GB"
},
{
"id": "starcoder:3b",
"parameters": "3B",
"size": "1.8GB"
},
{
"id": "starcoder:1b",
"parameters": "1B",
"size": "726MB"
}
],
"vision": false
},
{
"description": "Sentence-transformers model that can be used for tasks like clustering or semantic search.",
"id": "paraphrase-multilingual",
"name": "Paraphrase Multilingual",
"variants": [
{
"id": "paraphrase-multilingual:latest",
"parameters": "Unknown",
"size": "563MB"
},
{
"id": "paraphrase-multilingual:278m",
"parameters": "278M",
"size": "563MB"
}
],
"vision": false
},
{
"description": "General use models based on Llama and Llama 2 from Nous Research.",
"id": "nous-hermes",
"name": "Nous Hermes",
"variants": [
{
"id": "nous-hermes:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "nous-hermes:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "nous-hermes:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "EXAONE Deep exhibits superior capabilities in various reasoning tasks including math and coding benchmarks, ranging from 2.4B to 32B parameters developed and released by LG AI Research.",
"id": "exaone-deep",
"name": "Exaone Deep",
"variants": [
{
"id": "exaone-deep:latest",
"parameters": "Unknown",
"size": "4.8GB"
},
{
"id": "exaone-deep:32b",
"parameters": "32B",
"size": "19GB"
},
{
"id": "exaone-deep:7.8b",
"parameters": "7.8B",
"size": "4.8GB"
},
{
"id": "exaone-deep:2.4b",
"parameters": "2.4B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "An advanced language model crafted with 2 trillion bilingual tokens.",
"id": "deepseek-llm",
"name": "Deepseek Llm",
"variants": [
{
"id": "deepseek-llm:latest",
"parameters": "Unknown",
"size": "4.0GB"
},
{
"id": "deepseek-llm:67b",
"parameters": "67B",
"size": "38GB"
},
{
"id": "deepseek-llm:7b",
"parameters": "7B",
"size": "4.0GB"
}
],
"vision": false
},
{
"description": "A large language model built by the Technology Innovation Institute (TII) for use in summarization, text generation, and chat bots.",
"id": "falcon",
"name": "Falcon",
"variants": [
{
"id": "falcon:latest",
"parameters": "Unknown",
"size": "4.2GB"
},
{
"id": "falcon:180b",
"parameters": "180B",
"size": "101GB"
},
{
"id": "falcon:40b",
"parameters": "40B",
"size": "24GB"
},
{
"id": "falcon:7b",
"parameters": "7B",
"size": "4.2GB"
}
],
"vision": false
},
{
"description": "A strong, economical, and efficient Mixture-of-Experts language model.",
"id": "deepseek-v2",
"name": "Deepseek V2",
"variants": [
{
"id": "deepseek-v2:latest",
"parameters": "Unknown",
"size": "8.9GB"
},
{
"id": "deepseek-v2:236b",
"parameters": "236B",
"size": "133GB"
},
{
"id": "deepseek-v2:16b",
"parameters": "16B",
"size": "8.9GB"
}
],
"vision": false
},
{
"description": "A family of open-source models trained on a wide variety of data, surpassing ChatGPT on various benchmarks. Updated to version 3.5-0106.",
"id": "openchat",
"name": "Openchat",
"variants": [
{
"id": "openchat:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "openchat:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "General use chat model based on Llama and Llama 2 with 2K to 16K context sizes.",
"id": "vicuna",
"name": "Vicuna",
"variants": [
{
"id": "vicuna:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "vicuna:33b",
"parameters": "33B",
"size": "18GB"
},
{
"id": "vicuna:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "vicuna:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "The Ministral 3 family is designed for edge deployment, capable of running on a wide range of hardware.",
"id": "ministral-3",
"name": "Ministral 3",
"variants": [
{
"id": "ministral-3:latest",
"parameters": "Unknown",
"size": "6.0GB"
},
{
"id": "ministral-3:14b",
"parameters": "14B",
"size": "9.1GB"
},
{
"id": "ministral-3:14b-cloud",
"parameters": "14B",
"size": "Cloud (API only)"
},
{
"id": "ministral-3:8b",
"parameters": "8B",
"size": "6.0GB"
},
{
"id": "ministral-3:8b-cloud",
"parameters": "8B",
"size": "Cloud (API only)"
},
{
"id": "ministral-3:3b",
"parameters": "3B",
"size": "3.0GB"
},
{
"id": "ministral-3:3b-cloud",
"parameters": "3B",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"cloud": true
},
{
"description": "OpenHermes 2.5 is a 7B model fine-tuned by Teknium on Mistral with fully open datasets.",
"id": "openhermes",
"name": "Openhermes",
"variants": [
{
"id": "openhermes:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "openhermes:7b-mistral-v2-q2_k",
"parameters": "7B",
"size": "3.1GB"
},
{
"id": "openhermes:7b-mistral-v2-q3_k_s",
"parameters": "7B",
"size": "3.2GB"
},
{
"id": "openhermes:7b-mistral-v2-q3_k_m",
"parameters": "7B",
"size": "3.5GB"
},
{
"id": "openhermes:7b-mistral-v2-q3_k_l",
"parameters": "7B",
"size": "3.8GB"
},
{
"id": "openhermes:7b-mistral-v2-q4_0",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:7b-mistral-v2-q4_1",
"parameters": "7B",
"size": "4.6GB"
},
{
"id": "openhermes:7b-mistral-v2-q4_k_s",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:7b-mistral-v2-q4_k_m",
"parameters": "7B",
"size": "4.4GB"
},
{
"id": "openhermes:7b-mistral-v2-q5_0",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "openhermes:7b-mistral-v2-q5_1",
"parameters": "7B",
"size": "5.4GB"
},
{
"id": "openhermes:7b-mistral-v2-q5_k_s",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "openhermes:7b-mistral-v2-q5_k_m",
"parameters": "7B",
"size": "5.1GB"
},
{
"id": "openhermes:7b-mistral-v2-q6_k",
"parameters": "7B",
"size": "5.9GB"
},
{
"id": "openhermes:7b-mistral-v2-q8_0",
"parameters": "7B",
"size": "7.7GB"
},
{
"id": "openhermes:7b-mistral-v2-fp16",
"parameters": "7B",
"size": "14GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q2_k",
"parameters": "7B",
"size": "3.1GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q3_k_s",
"parameters": "7B",
"size": "3.2GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q3_k_m",
"parameters": "7B",
"size": "3.5GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q3_k_l",
"parameters": "7B",
"size": "3.8GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q4_0",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q4_1",
"parameters": "7B",
"size": "4.6GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q4_k_s",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q4_k_m",
"parameters": "7B",
"size": "4.4GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q5_0",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q5_1",
"parameters": "7B",
"size": "5.4GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q5_k_s",
"parameters": "7B",
"size": "5.0GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q5_k_m",
"parameters": "7B",
"size": "5.1GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q6_k",
"parameters": "7B",
"size": "5.9GB"
},
{
"id": "openhermes:7b-mistral-v2.5-q8_0",
"parameters": "7B",
"size": "7.7GB"
},
{
"id": "openhermes:7b-mistral-v2.5-fp16",
"parameters": "7B",
"size": "14GB"
},
{
"id": "openhermes:7b-v2",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:7b-v2.5",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "openhermes:v2",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "openhermes:v2.5",
"parameters": "Unknown",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "CodeQwen1.5 is a large language model pretrained on a large amount of code data.",
"id": "codeqwen",
"name": "Codeqwen",
"variants": [
{
"id": "codeqwen:latest",
"parameters": "Unknown",
"size": "4.2GB"
},
{
"id": "codeqwen:7b",
"parameters": "7B",
"size": "4.2GB"
}
],
"vision": false
},
{
"description": "A strong multi-lingual general language model with competitive performance to Llama 3.",
"id": "glm4",
"name": "Glm4",
"variants": [
{
"id": "glm4:latest",
"parameters": "Unknown",
"size": "5.5GB"
},
{
"id": "glm4:9b",
"parameters": "9B",
"size": "5.5GB"
}
],
"vision": false
},
{
"description": "Qwen2 Math is a series of specialized math language models built upon the Qwen2 LLMs, which significantly outperforms the mathematical capabilities of open-source models and even closed-source models (e.g., GPT4o).",
"id": "qwen2-math",
"name": "Qwen2 Math",
"variants": [
{
"id": "qwen2-math:latest",
"parameters": "Unknown",
"size": "4.4GB"
},
{
"id": "qwen2-math:72b",
"parameters": "72B",
"size": "41GB"
},
{
"id": "qwen2-math:7b",
"parameters": "7B",
"size": "4.4GB"
},
{
"id": "qwen2-math:1.5b",
"parameters": "1.5B",
"size": "935MB"
}
],
"vision": false
},
{
"description": "Aya 23, released by Cohere, is a new family of state-of-the-art, multilingual models that support 23 languages. ",
"id": "aya",
"name": "Aya",
"variants": [
{
"id": "aya:latest",
"parameters": "Unknown",
"size": "4.8GB"
},
{
"id": "aya:35b",
"parameters": "35B",
"size": "20GB"
},
{
"id": "aya:8b",
"parameters": "8B",
"size": "4.8GB"
}
],
"vision": false
},
{
"description": "Llama 2 based model fine tuned to improve Chinese dialogue ability.",
"id": "llama2-chinese",
"name": "Llama2 Chinese",
"variants": [
{
"id": "llama2-chinese:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "llama2-chinese:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "llama2-chinese:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Stable Code 3B is a coding model with instruct and code completion variants on par with models such as Code Llama 7B that are 2.5x larger.",
"id": "stable-code",
"name": "Stable Code",
"variants": [
{
"id": "stable-code:latest",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "stable-code:3b",
"parameters": "3B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "A fine-tuned model based on Mistral with good coverage of domain and language.",
"id": "neural-chat",
"name": "Neural Chat",
"variants": [
{
"id": "neural-chat:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "neural-chat:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "The powerful family of models by Nous Research that excels at scientific discussion and coding tasks.",
"id": "nous-hermes2",
"name": "Nous Hermes2",
"variants": [
{
"id": "nous-hermes2:latest",
"parameters": "Unknown",
"size": "6.1GB"
},
{
"id": "nous-hermes2:34b",
"parameters": "34B",
"size": "19GB"
},
{
"id": "nous-hermes2:10.7b",
"parameters": "10.7B",
"size": "6.1GB"
}
],
"vision": false
},
{
"description": "OpenCoder is an open and reproducible code LLM family which includes 1.5B and 8B models, supporting chat in English and Chinese languages.",
"id": "opencoder",
"name": "Opencoder",
"variants": [
{
"id": "opencoder:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "opencoder:8b",
"parameters": "8B",
"size": "4.7GB"
},
{
"id": "opencoder:1.5b",
"parameters": "1.5B",
"size": "1.4GB"
}
],
"vision": false
},
{
"description": "SQLCoder is a code completion model fined-tuned on StarCoder for SQL generation tasks",
"id": "sqlcoder",
"name": "Sqlcoder",
"variants": [
{
"id": "sqlcoder:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "sqlcoder:15b",
"parameters": "15B",
"size": "9.0GB"
},
{
"id": "sqlcoder:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "State-of-the-art code generation model",
"id": "wizardcoder",
"name": "Wizardcoder",
"variants": [
{
"id": "wizardcoder:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "wizardcoder:33b",
"parameters": "33B",
"size": "19GB"
}
],
"vision": false
},
{
"description": "Yi-Coder is a series of open-source code language models that delivers state-of-the-art coding performance with fewer than 10 billion parameters.",
"id": "yi-coder",
"name": "Yi Coder",
"variants": [
{
"id": "yi-coder:latest",
"parameters": "Unknown",
"size": "5.0GB"
},
{
"id": "yi-coder:9b",
"parameters": "9B",
"size": "5.0GB"
},
{
"id": "yi-coder:1.5b",
"parameters": "1.5B",
"size": "866MB"
}
],
"vision": false
},
{
"description": "Stable LM 2 is a state-of-the-art 1.6B and 12B parameter language model trained on multilingual data in English, Spanish, German, Italian, French, Portuguese, and Dutch.",
"id": "stablelm2",
"name": "Stablelm2",
"variants": [
{
"id": "stablelm2:latest",
"parameters": "Unknown",
"size": "983MB"
},
{
"id": "stablelm2:12b",
"parameters": "12B",
"size": "7.0GB"
},
{
"id": "stablelm2:1.6b",
"parameters": "1.6B",
"size": "983MB"
}
],
"vision": false
},
{
"description": "A model from NVIDIA based on Llama 3 that excels at conversational question answering (QA) and retrieval-augmented generation (RAG).",
"id": "llama3-chatqa",
"name": "Llama3 Chatqa",
"variants": [
{
"id": "llama3-chatqa:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama3-chatqa:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "llama3-chatqa:8b",
"parameters": "8B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "The IBM Granite 2B and 8B models are designed to support tool-based use cases and support for retrieval augmented generation (RAG), streamlining code generation, translation and bug fixing.",
"id": "granite3-dense",
"name": "Granite3 Dense",
"variants": [
{
"id": "granite3-dense:latest",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "granite3-dense:8b",
"parameters": "8B",
"size": "4.9GB"
},
{
"id": "granite3-dense:2b",
"parameters": "2B",
"size": "1.6GB"
}
],
"vision": false,
"tools": true
},
{
"description": "The IBM Granite 2B and 8B models are text-only dense LLMs trained on over 12 trillion tokens of data, demonstrated significant improvements over their predecessors in performance and speed in IBM’s initial testing.",
"id": "granite3.1-dense",
"name": "Granite3.1 Dense",
"variants": [
{
"id": "granite3.1-dense:latest",
"parameters": "Unknown",
"size": "5.0GB"
},
{
"id": "granite3.1-dense:8b",
"parameters": "8B",
"size": "5.0GB"
},
{
"id": "granite3.1-dense:2b",
"parameters": "2B",
"size": "1.6GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Model focused on math and logic problems",
"id": "wizard-math",
"name": "Wizard Math",
"variants": [
{
"id": "wizard-math:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "wizard-math:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "wizard-math:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizard-math:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "A 7B and 15B uncensored variant of the Dolphin model family that excels at coding, based on StarCoder2.",
"id": "dolphincoder",
"name": "Dolphincoder",
"variants": [
{
"id": "dolphincoder:latest",
"parameters": "Unknown",
"size": "4.2GB"
},
{
"id": "dolphincoder:15b",
"parameters": "15B",
"size": "9.1GB"
},
{
"id": "dolphincoder:7b",
"parameters": "7B",
"size": "4.2GB"
}
],
"vision": false
},
{
"description": "This model extends LLama-3 8B's context length from 8k to over 1m tokens.",
"id": "llama3-gradient",
"name": "Llama3 Gradient",
"variants": [
{
"id": "llama3-gradient:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama3-gradient:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "llama3-gradient:8b",
"parameters": "8B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "A companion assistant trained in philosophy, psychology, and personal relationships. Based on Mistral.",
"id": "samantha-mistral",
"name": "Samantha Mistral",
"variants": [
{
"id": "samantha-mistral:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "samantha-mistral:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "DeepSeek-V3.1-Terminus is a hybrid model that supports both thinking mode and non-thinking mode.",
"id": "deepseek-v3.1",
"name": "Deepseek V3.1",
"variants": [
{
"id": "deepseek-v3.1:latest",
"parameters": "Unknown",
"size": "404GB"
},
{
"id": "deepseek-v3.1:671b",
"parameters": "671B",
"size": "404GB"
},
{
"id": "deepseek-v3.1:671b-cloud",
"parameters": "671B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Command R+ is a powerful, scalable large language model purpose-built to excel at real-world enterprise use cases.",
"id": "command-r-plus",
"name": "Command R Plus",
"variants": [
{
"id": "command-r-plus:latest",
"parameters": "Unknown",
"size": "59GB"
},
{
"id": "command-r-plus:104b",
"parameters": "104B",
"size": "59GB"
}
],
"vision": false,
"tools": true
},
{
"description": "InternLM2.5 is a 7B parameter model tailored for practical scenarios with outstanding reasoning capability.",
"id": "internlm2",
"name": "Internlm2",
"variants": [
{
"id": "internlm2:latest",
"parameters": "Unknown",
"size": "4.5GB"
},
{
"id": "internlm2:20b",
"parameters": "20B",
"size": "11GB"
},
{
"id": "internlm2:7b",
"parameters": "7B",
"size": "4.5GB"
},
{
"id": "internlm2:1.8b",
"parameters": "1.8B",
"size": "1.1GB"
},
{
"id": "internlm2:1m",
"parameters": "1M",
"size": "4.5GB"
}
],
"vision": false
},
{
"description": "A series of models from Groq that represent a significant advancement in open-source AI capabilities for tool use/function calling.",
"id": "llama3-groq-tool-use",
"name": "Llama3 Groq Tool Use",
"variants": [
{
"id": "llama3-groq-tool-use:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama3-groq-tool-use:70b",
"parameters": "70B",
"size": "40GB"
},
{
"id": "llama3-groq-tool-use:8b",
"parameters": "8B",
"size": "4.7GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Llama Guard 3 is a series of models fine-tuned for content safety classification of LLM inputs and responses.",
"id": "llama-guard3",
"name": "Llama Guard3",
"variants": [
{
"id": "llama-guard3:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "llama-guard3:8b",
"parameters": "8B",
"size": "4.9GB"
},
{
"id": "llama-guard3:1b",
"parameters": "1B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "Starling is a large language model trained by reinforcement learning from AI feedback focused on improving chatbot helpfulness.",
"id": "starling-lm",
"name": "Starling Lm",
"variants": [
{
"id": "starling-lm:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "starling-lm:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Code generation model based on Code Llama.",
"id": "phind-codellama",
"name": "Phind Codellama",
"variants": [
{
"id": "phind-codellama:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "phind-codellama:34b",
"parameters": "34B",
"size": "19GB"
}
],
"vision": false
},
{
"description": "A compact, yet powerful 10.7B large language model designed for single-turn conversation.",
"id": "solar",
"name": "Solar",
"variants": [
{
"id": "solar:latest",
"parameters": "Unknown",
"size": "6.1GB"
},
{
"id": "solar:10.7b",
"parameters": "10.7B",
"size": "6.1GB"
}
],
"vision": false
},
{
"description": "Conversational model based on Llama 2 that performs competitively on various benchmarks.",
"id": "xwinlm",
"name": "Xwinlm",
"variants": [
{
"id": "xwinlm:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "xwinlm:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "xwinlm:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Cohere For AI's language models trained to perform well across 23 different languages.",
"id": "aya-expanse",
"name": "Aya Expanse",
"variants": [
{
"id": "aya-expanse:latest",
"parameters": "Unknown",
"size": "5.1GB"
},
{
"id": "aya-expanse:32b",
"parameters": "32B",
"size": "20GB"
},
{
"id": "aya-expanse:8b",
"parameters": "8B",
"size": "5.1GB"
}
],
"vision": false,
"tools": true
},
{
"description": "The IBM Granite 1B and 3B models are the first mixture of experts (MoE) Granite models from IBM designed for low latency usage.",
"id": "granite3-moe",
"name": "Granite3 Moe",
"variants": [
{
"id": "granite3-moe:latest",
"parameters": "Unknown",
"size": "822MB"
},
{
"id": "granite3-moe:3b",
"parameters": "3B",
"size": "2.1GB"
},
{
"id": "granite3-moe:1b",
"parameters": "1B",
"size": "822MB"
}
],
"vision": false,
"tools": true
},
{
"description": "An extension of Llama 2 that supports a context of up to 128k tokens.",
"id": "yarn-llama2",
"name": "Yarn Llama2",
"variants": [
{
"id": "yarn-llama2:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "yarn-llama2:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "yarn-llama2:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "A new collection of open translation models built on Gemma 3, helping people communicate across 55 languages.",
"id": "translategemma",
"name": "Translategemma",
"variants": [
{
"id": "translategemma:latest",
"parameters": "Unknown",
"size": "3.3GB"
},
{
"id": "translategemma:27b",
"parameters": "27B",
"size": "17GB"
},
{
"id": "translategemma:12b",
"parameters": "12B",
"size": "8.1GB"
},
{
"id": "translategemma:4b",
"parameters": "4B",
"size": "3.3GB"
}
],
"vision": true
},
{
"description": "A versatile model for AI software development scenarios, including code completion.",
"id": "codegeex4",
"name": "Codegeex4",
"variants": [
{
"id": "codegeex4:latest",
"parameters": "Unknown",
"size": "5.5GB"
},
{
"id": "codegeex4:9b",
"parameters": "9B",
"size": "5.5GB"
}
],
"vision": false
},
{
"description": "Mistral OpenOrca is a 7 billion parameter model, fine-tuned on top of the Mistral 7B model using the OpenOrca dataset.",
"id": "mistral-openorca",
"name": "Mistral Openorca",
"variants": [
{
"id": "mistral-openorca:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "mistral-openorca:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "An experimental 1.1B parameter model trained on the new Dolphin 2.8 dataset by Eric Hartford and based on TinyLlama.",
"id": "tinydolphin",
"name": "Tinydolphin",
"variants": [
{
"id": "tinydolphin:latest",
"parameters": "Unknown",
"size": "637MB"
},
{
"id": "tinydolphin:1.1b",
"parameters": "1.1B",
"size": "637MB"
}
],
"vision": false
},
{
"description": "Orca 2 is built by Microsoft research, and are a fine-tuned version of Meta's Llama 2 models. The model is designed to excel particularly in reasoning.",
"id": "orca2",
"name": "Orca2",
"variants": [
{
"id": "orca2:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "orca2:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "orca2:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "The first installment in the Qwen3-Next series with strong performance in terms of both parameter efficiency and inference speed.",
"id": "qwen3-next",
"name": "Qwen3 Next",
"variants": [
{
"id": "qwen3-next:latest",
"parameters": "Unknown",
"size": "50GB"
},
{
"id": "qwen3-next:80b",
"parameters": "80B",
"size": "50GB"
},
{
"id": "qwen3-next:80b-cloud",
"parameters": "80B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Llama 2 based model fine tuned on an Orca-style dataset. Originally called Free Willy.",
"id": "stable-beluga",
"name": "Stable Beluga",
"variants": [
{
"id": "stable-beluga:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "stable-beluga:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "stable-beluga:13b",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "stable-beluga:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "A series of models that convert HTML content to Markdown content, which is useful for content conversion tasks.",
"id": "reader-lm",
"name": "Reader Lm",
"variants": [
{
"id": "reader-lm:latest",
"parameters": "Unknown",
"size": "935MB"
},
{
"id": "reader-lm:1.5b",
"parameters": "1.5B",
"size": "935MB"
},
{
"id": "reader-lm:0.5b",
"parameters": "0.5B",
"size": "352MB"
}
],
"vision": false
},
{
"description": "ShieldGemma is set of instruction tuned models for evaluating the safety of text prompt input and text output responses against a set of defined safety policies.",
"id": "shieldgemma",
"name": "Shieldgemma",
"variants": [
{
"id": "shieldgemma:latest",
"parameters": "Unknown",
"size": "5.8GB"
},
{
"id": "shieldgemma:27b",
"parameters": "27B",
"size": "17GB"
},
{
"id": "shieldgemma:9b",
"parameters": "9B",
"size": "5.8GB"
},
{
"id": "shieldgemma:2b",
"parameters": "2B",
"size": "1.7GB"
}
],
"vision": false
},
{
"description": "Rnj-1 is a family of 8B parameter open-weight, dense models trained from scratch by Essential AI, optimized for code and STEM with capabilities on par with SOTA open-weight models.",
"id": "rnj-1",
"name": "Rnj 1",
"variants": [
{
"id": "rnj-1:latest",
"parameters": "Unknown",
"size": "5.1GB"
},
{
"id": "rnj-1:8b",
"parameters": "8B",
"size": "5.1GB"
},
{
"id": "rnj-1:8b-cloud",
"parameters": "8B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "An expansion of Llama 2 that specializes in integrating both general language understanding and domain-specific knowledge, particularly in programming and mathematics.",
"id": "llama-pro",
"name": "Llama Pro",
"variants": [
{
"id": "llama-pro:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama-pro:8b-instruct-q2_k",
"parameters": "8B",
"size": "3.5GB"
},
{
"id": "llama-pro:8b-instruct-q3_k_s",
"parameters": "8B",
"size": "3.6GB"
},
{
"id": "llama-pro:8b-instruct-q3_k_m",
"parameters": "8B",
"size": "4.1GB"
},
{
"id": "llama-pro:8b-instruct-q3_k_l",
"parameters": "8B",
"size": "4.5GB"
},
{
"id": "llama-pro:8b-instruct-q4_0",
"parameters": "8B",
"size": "4.7GB"
},
{
"id": "llama-pro:8b-instruct-q4_1",
"parameters": "8B",
"size": "5.3GB"
},
{
"id": "llama-pro:8b-instruct-q4_k_s",
"parameters": "8B",
"size": "4.8GB"
},
{
"id": "llama-pro:8b-instruct-q4_k_m",
"parameters": "8B",
"size": "5.1GB"
},
{
"id": "llama-pro:8b-instruct-q5_0",
"parameters": "8B",
"size": "5.8GB"
},
{
"id": "llama-pro:8b-instruct-q5_1",
"parameters": "8B",
"size": "6.3GB"
},
{
"id": "llama-pro:8b-instruct-q5_k_s",
"parameters": "8B",
"size": "5.8GB"
},
{
"id": "llama-pro:8b-instruct-q5_k_m",
"parameters": "8B",
"size": "5.9GB"
},
{
"id": "llama-pro:8b-instruct-q6_k",
"parameters": "8B",
"size": "6.9GB"
},
{
"id": "llama-pro:8b-instruct-q8_0",
"parameters": "8B",
"size": "8.9GB"
},
{
"id": "llama-pro:8b-instruct-fp16",
"parameters": "8B",
"size": "17GB"
},
{
"id": "llama-pro:8b-text-q2_k",
"parameters": "8B",
"size": "3.5GB"
},
{
"id": "llama-pro:8b-text-q3_k_s",
"parameters": "8B",
"size": "3.6GB"
},
{
"id": "llama-pro:8b-text-q3_k_m",
"parameters": "8B",
"size": "4.1GB"
},
{
"id": "llama-pro:8b-text-q3_k_l",
"parameters": "8B",
"size": "4.5GB"
},
{
"id": "llama-pro:8b-text-q4_0",
"parameters": "8B",
"size": "4.7GB"
},
{
"id": "llama-pro:8b-text-q4_1",
"parameters": "8B",
"size": "5.3GB"
},
{
"id": "llama-pro:8b-text-q4_k_s",
"parameters": "8B",
"size": "4.8GB"
},
{
"id": "llama-pro:8b-text-q4_k_m",
"parameters": "8B",
"size": "5.1GB"
},
{
"id": "llama-pro:8b-text-q5_0",
"parameters": "8B",
"size": "5.8GB"
},
{
"id": "llama-pro:8b-text-q5_1",
"parameters": "8B",
"size": "6.3GB"
},
{
"id": "llama-pro:8b-text-q5_k_s",
"parameters": "8B",
"size": "5.8GB"
},
{
"id": "llama-pro:8b-text-q5_k_m",
"parameters": "8B",
"size": "5.9GB"
},
{
"id": "llama-pro:8b-text-q6_k",
"parameters": "8B",
"size": "6.9GB"
},
{
"id": "llama-pro:8b-text-q8_0",
"parameters": "8B",
"size": "8.9GB"
},
{
"id": "llama-pro:8b-text-fp16",
"parameters": "8B",
"size": "17GB"
},
{
"id": "llama-pro:instruct",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "llama-pro:text",
"parameters": "Unknown",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "An extension of Mistral to support context windows of 64K or 128K.",
"id": "yarn-mistral",
"name": "Yarn Mistral",
"variants": [
{
"id": "yarn-mistral:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "yarn-mistral:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Nexus Raven is a 13B instruction tuned model for function calling tasks. ",
"id": "nexusraven",
"name": "Nexusraven",
"variants": [
{
"id": "nexusraven:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "nexusraven:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "General use model based on Llama 2.",
"id": "wizardlm",
"name": "Wizardlm",
"variants": [
{
"id": "wizardlm:70b-llama2-q2_k",
"parameters": "70B",
"size": "29GB"
},
{
"id": "wizardlm:70b-llama2-q3_k_s",
"parameters": "70B",
"size": "30GB"
},
{
"id": "wizardlm:70b-llama2-q3_k_m",
"parameters": "70B",
"size": "33GB"
},
{
"id": "wizardlm:70b-llama2-q3_k_l",
"parameters": "70B",
"size": "36GB"
},
{
"id": "wizardlm:70b-llama2-q4_0",
"parameters": "70B",
"size": "39GB"
},
{
"id": "wizardlm:70b-llama2-q4_1",
"parameters": "70B",
"size": "43GB"
},
{
"id": "wizardlm:70b-llama2-q4_k_s",
"parameters": "70B",
"size": "39GB"
},
{
"id": "wizardlm:70b-llama2-q4_k_m",
"parameters": "70B",
"size": "41GB"
},
{
"id": "wizardlm:70b-llama2-q5_0",
"parameters": "70B",
"size": "47GB"
},
{
"id": "wizardlm:70b-llama2-q5_k_s",
"parameters": "70B",
"size": "47GB"
},
{
"id": "wizardlm:70b-llama2-q5_k_m",
"parameters": "70B",
"size": "49GB"
},
{
"id": "wizardlm:70b-llama2-q6_k",
"parameters": "70B",
"size": "57GB"
},
{
"id": "wizardlm:70b-llama2-q8_0",
"parameters": "70B",
"size": "73GB"
},
{
"id": "wizardlm:30b-q2_k",
"parameters": "30B",
"size": "14GB"
},
{
"id": "wizardlm:30b-q3_k_s",
"parameters": "30B",
"size": "14GB"
},
{
"id": "wizardlm:30b-q3_k_m",
"parameters": "30B",
"size": "16GB"
},
{
"id": "wizardlm:30b-q3_k_l",
"parameters": "30B",
"size": "17GB"
},
{
"id": "wizardlm:30b-q4_0",
"parameters": "30B",
"size": "18GB"
},
{
"id": "wizardlm:30b-q4_1",
"parameters": "30B",
"size": "20GB"
},
{
"id": "wizardlm:30b-q4_k_s",
"parameters": "30B",
"size": "18GB"
},
{
"id": "wizardlm:30b-q4_k_m",
"parameters": "30B",
"size": "20GB"
},
{
"id": "wizardlm:30b-q5_0",
"parameters": "30B",
"size": "22GB"
},
{
"id": "wizardlm:30b-q5_1",
"parameters": "30B",
"size": "24GB"
},
{
"id": "wizardlm:30b-q5_k_s",
"parameters": "30B",
"size": "22GB"
},
{
"id": "wizardlm:30b-q5_k_m",
"parameters": "30B",
"size": "23GB"
},
{
"id": "wizardlm:30b-q6_k",
"parameters": "30B",
"size": "27GB"
},
{
"id": "wizardlm:30b-q8_0",
"parameters": "30B",
"size": "35GB"
},
{
"id": "wizardlm:30b-fp16",
"parameters": "30B",
"size": "65GB"
},
{
"id": "wizardlm:13b-llama2-q2_k",
"parameters": "13B",
"size": "5.4GB"
},
{
"id": "wizardlm:13b-llama2-q3_k_s",
"parameters": "13B",
"size": "5.7GB"
},
{
"id": "wizardlm:13b-llama2-q3_k_m",
"parameters": "13B",
"size": "6.3GB"
},
{
"id": "wizardlm:13b-llama2-q3_k_l",
"parameters": "13B",
"size": "6.9GB"
},
{
"id": "wizardlm:13b-llama2-q4_0",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizardlm:13b-llama2-q4_1",
"parameters": "13B",
"size": "8.2GB"
},
{
"id": "wizardlm:13b-llama2-q4_k_s",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizardlm:13b-llama2-q4_k_m",
"parameters": "13B",
"size": "7.9GB"
},
{
"id": "wizardlm:13b-llama2-q5_0",
"parameters": "13B",
"size": "9.0GB"
},
{
"id": "wizardlm:13b-llama2-q5_1",
"parameters": "13B",
"size": "9.8GB"
},
{
"id": "wizardlm:13b-llama2-q5_k_s",
"parameters": "13B",
"size": "9.0GB"
},
{
"id": "wizardlm:13b-llama2-q5_k_m",
"parameters": "13B",
"size": "9.2GB"
},
{
"id": "wizardlm:13b-llama2-q6_k",
"parameters": "13B",
"size": "11GB"
},
{
"id": "wizardlm:13b-llama2-q8_0",
"parameters": "13B",
"size": "14GB"
},
{
"id": "wizardlm:13b-llama2-fp16",
"parameters": "13B",
"size": "26GB"
},
{
"id": "wizardlm:13b-q2_k",
"parameters": "13B",
"size": "5.4GB"
},
{
"id": "wizardlm:13b-q3_k_s",
"parameters": "13B",
"size": "5.7GB"
},
{
"id": "wizardlm:13b-q3_k_m",
"parameters": "13B",
"size": "6.3GB"
},
{
"id": "wizardlm:13b-q3_k_l",
"parameters": "13B",
"size": "6.9GB"
},
{
"id": "wizardlm:13b-q4_0",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizardlm:13b-q4_1",
"parameters": "13B",
"size": "8.2GB"
},
{
"id": "wizardlm:13b-q4_k_s",
"parameters": "13B",
"size": "7.4GB"
},
{
"id": "wizardlm:13b-q4_k_m",
"parameters": "13B",
"size": "7.9GB"
},
{
"id": "wizardlm:13b-q5_0",
"parameters": "13B",
"size": "9.0GB"
},
{
"id": "wizardlm:13b-q5_1",
"parameters": "13B",
"size": "9.8GB"
},
{
"id": "wizardlm:13b-q5_k_s",
"parameters": "13B",
"size": "9.0GB"
},
{
"id": "wizardlm:13b-q5_k_m",
"parameters": "13B",
"size": "9.2GB"
},
{
"id": "wizardlm:13b-q6_k",
"parameters": "13B",
"size": "11GB"
},
{
"id": "wizardlm:13b-q8_0",
"parameters": "13B",
"size": "14GB"
},
{
"id": "wizardlm:13b-fp16",
"parameters": "13B",
"size": "26GB"
},
{
"id": "wizardlm:7b-q2_k",
"parameters": "7B",
"size": "2.8GB"
},
{
"id": "wizardlm:7b-q3_k_s",
"parameters": "7B",
"size": "2.9GB"
},
{
"id": "wizardlm:7b-q3_k_m",
"parameters": "7B",
"size": "3.3GB"
},
{
"id": "wizardlm:7b-q3_k_l",
"parameters": "7B",
"size": "3.6GB"
},
{
"id": "wizardlm:7b-q4_0",
"parameters": "7B",
"size": "3.8GB"
},
{
"id": "wizardlm:7b-q4_1",
"parameters": "7B",
"size": "4.2GB"
},
{
"id": "wizardlm:7b-q4_k_s",
"parameters": "7B",
"size": "3.9GB"
},
{
"id": "wizardlm:7b-q4_k_m",
"parameters": "7B",
"size": "4.1GB"
},
{
"id": "wizardlm:7b-q5_0",
"parameters": "7B",
"size": "4.7GB"
},
{
"id": "wizardlm:7b-q5_1",
"parameters": "7B",
"size": "5.1GB"
},
{
"id": "wizardlm:7b-q5_k_s",
"parameters": "7B",
"size": "4.7GB"
},
{
"id": "wizardlm:7b-q5_k_m",
"parameters": "7B",
"size": "4.8GB"
},
{
"id": "wizardlm:7b-q6_k",
"parameters": "7B",
"size": "5.5GB"
},
{
"id": "wizardlm:7b-q8_0",
"parameters": "7B",
"size": "7.2GB"
},
{
"id": "wizardlm:7b-fp16",
"parameters": "7B",
"size": "13GB"
}
],
"vision": false
},
{
"description": "Open-source medical large language model adapted from Llama 2 to the medical domain.",
"id": "meditron",
"name": "Meditron",
"variants": [
{
"id": "meditron:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "meditron:70b",
"parameters": "70B",
"size": "39GB"
},
{
"id": "meditron:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "A high-performing model trained with a new technique called Reflection-tuning that teaches a LLM to detect mistakes in its reasoning and correct course.",
"id": "reflection",
"name": "Reflection",
"variants": [
{
"id": "reflection:latest",
"parameters": "Unknown",
"size": "40GB"
},
{
"id": "reflection:70b",
"parameters": "70B",
"size": "40GB"
}
],
"vision": false
},
{
"description": "A commercial-friendly small language model by NVIDIA optimized for roleplay, RAG QA, and function calling.",
"id": "nemotron-mini",
"name": "Nemotron Mini",
"variants": [
{
"id": "nemotron-mini:latest",
"parameters": "Unknown",
"size": "2.7GB"
},
{
"id": "nemotron-mini:4b",
"parameters": "4B",
"size": "2.7GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Granite-3.2 is a family of long-context AI models from IBM Granite fine-tuned for thinking capabilities.",
"id": "granite3.2",
"name": "Granite3.2",
"variants": [
{
"id": "granite3.2:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "granite3.2:8b",
"parameters": "8B",
"size": "4.9GB"
},
{
"id": "granite3.2:2b",
"parameters": "2B",
"size": "1.5GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "As the strongest model in the 30B class, GLM-4.7-Flash offers a new option for lightweight deployment that balances performance and efficiency.",
"id": "glm-4.7-flash",
"name": "Glm 4.7 Flash",
"variants": [
{
"id": "glm-4.7-flash:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "glm-4.7-flash:q4_k_m",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "glm-4.7-flash:q8_0",
"parameters": "Unknown",
"size": "32GB"
},
{
"id": "glm-4.7-flash:bf16",
"parameters": "Unknown",
"size": "60GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "Uncensored version of Wizard LM model ",
"id": "wizardlm-uncensored",
"name": "Wizardlm Uncensored",
"variants": [
{
"id": "wizardlm-uncensored:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "wizardlm-uncensored:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "Athene-V2 is a 72B parameter model which excels at code completion, mathematics, and log extraction tasks.",
"id": "athene-v2",
"name": "Athene V2",
"variants": [
{
"id": "athene-v2:latest",
"parameters": "Unknown",
"size": "47GB"
},
{
"id": "athene-v2:72b",
"parameters": "72B",
"size": "47GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Llama-3.1-Nemotron-70B-Instruct is a large language model customized by NVIDIA to improve the helpfulness of LLM generated responses to user queries.",
"id": "nemotron",
"name": "Nemotron",
"variants": [
{
"id": "nemotron:latest",
"parameters": "Unknown",
"size": "43GB"
},
{
"id": "nemotron:70b",
"parameters": "70B",
"size": "43GB"
}
],
"vision": false,
"tools": true
},
{
"description": "EXAONE 3.5 is a collection of instruction-tuned bilingual (English and Korean) generative models ranging from 2.4B to 32B parameters, developed and released by LG AI Research. ",
"id": "exaone3.5",
"name": "Exaone3.5",
"variants": [
{
"id": "exaone3.5:latest",
"parameters": "Unknown",
"size": "4.8GB"
},
{
"id": "exaone3.5:32b",
"parameters": "32B",
"size": "19GB"
},
{
"id": "exaone3.5:7.8b",
"parameters": "7.8B",
"size": "4.8GB"
},
{
"id": "exaone3.5:2.4b",
"parameters": "2.4B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "Snowflake's frontier embedding model. Arctic Embed 2.0 adds multilingual support without sacrificing English performance or scalability.",
"id": "snowflake-arctic-embed2",
"name": "Snowflake Arctic Embed2",
"variants": [
{
"id": "snowflake-arctic-embed2:latest",
"parameters": "Unknown",
"size": "1.2GB"
},
{
"id": "snowflake-arctic-embed2:568m",
"parameters": "568M",
"size": "1.2GB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "The Nous Hermes 2 model from Nous Research, now trained over Mixtral.",
"id": "nous-hermes2-mixtral",
"name": "Nous Hermes2 Mixtral",
"variants": [
{
"id": "nous-hermes2-mixtral:latest",
"parameters": "Unknown",
"size": "26GB"
},
{
"id": "nous-hermes2-mixtral:8x7b",
"parameters": "7B",
"size": "26GB"
}
],
"vision": false
},
{
"description": "A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity. ",
"id": "r1-1776",
"name": "R1.1776",
"variants": [
{
"id": "r1-1776:latest",
"parameters": "Unknown",
"size": "43GB"
},
{
"id": "r1-1776:671b",
"parameters": "671B",
"size": "404GB"
},
{
"id": "r1-1776:70b",
"parameters": "70B",
"size": "43GB"
}
],
"vision": false
},
{
"description": "Fine-tuned Llama 2 model to answer medical questions based on an open source medical dataset. ",
"id": "medllama2",
"name": "Medllama2",
"variants": [
{
"id": "medllama2:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "medllama2:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Great code generation model based on Llama2.",
"id": "codeup",
"name": "Codeup",
"variants": [
{
"id": "codeup:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "codeup:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "Uncensored Llama2 based model with support for a 16K context window.",
"id": "everythinglm",
"name": "Everythinglm",
"variants": [
{
"id": "everythinglm:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "everythinglm:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "MathΣtral: a 7B model designed for math reasoning and scientific discovery by Mistral AI.",
"id": "mathstral",
"name": "Mathstral",
"variants": [
{
"id": "mathstral:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "mathstral:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Solar Pro Preview: an advanced large language model (LLM) with 22 billion parameters designed to fit into a single GPU",
"id": "solar-pro",
"name": "Solar Pro",
"variants": [
{
"id": "solar-pro:latest",
"parameters": "Unknown",
"size": "13GB"
},
{
"id": "solar-pro:22b",
"parameters": "22B",
"size": "13GB"
}
],
"vision": false
},
{
"description": "🎩 Magicoder is a family of 7B parameter models trained on 75K synthetic instruction data using OSS-Instruct, a novel approach to enlightening LLMs with open-source code snippets.",
"id": "magicoder",
"name": "Magicoder",
"variants": [
{
"id": "magicoder:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "magicoder:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Falcon2 is an 11B parameters causal decoder-only model built by TII and trained over 5T tokens.",
"id": "falcon2",
"name": "Falcon2",
"variants": [
{
"id": "falcon2:latest",
"parameters": "Unknown",
"size": "6.4GB"
},
{
"id": "falcon2:11b",
"parameters": "11B",
"size": "6.4GB"
}
],
"vision": false
},
{
"description": "A lightweight chat model allowing accurate, and responsive output without requiring high-end hardware.",
"id": "stablelm-zephyr",
"name": "Stablelm Zephyr",
"variants": [
{
"id": "stablelm-zephyr:latest",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "stablelm-zephyr:3b",
"parameters": "3B",
"size": "1.6GB"
}
],
"vision": false
},
{
"description": "MegaDolphin-2.2-120b is a transformation of Dolphin-2.2-70b created by interleaving the model with itself.",
"id": "megadolphin",
"name": "Megadolphin",
"variants": [
{
"id": "megadolphin:latest",
"parameters": "Unknown",
"size": "68GB"
},
{
"id": "megadolphin:120b",
"parameters": "120B",
"size": "68GB"
}
],
"vision": false
},
{
"description": "The IBM Granite Embedding 30M and 278M models models are text-only dense biencoder embedding models, with 30M available in English only and 278M serving multilingual use cases.",
"id": "granite-embedding",
"name": "Granite Embedding",
"variants": [
{
"id": "granite-embedding:latest",
"parameters": "Unknown",
"size": "63MB"
},
{
"id": "granite-embedding:278m",
"parameters": "278M",
"size": "563MB"
},
{
"id": "granite-embedding:30m",
"parameters": "30M",
"size": "63MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "7B parameter text-to-SQL model made by MotherDuck and Numbers Station.",
"id": "duckdb-nsql",
"name": "Duckdb Nsql",
"variants": [
{
"id": "duckdb-nsql:latest",
"parameters": "Unknown",
"size": "3.8GB"
},
{
"id": "duckdb-nsql:7b",
"parameters": "7B",
"size": "3.8GB"
}
],
"vision": false
},
{
"description": "Tülu 3 is a leading instruction following model family, offering fully open-source data, code, and recipes by the The Allen Institute for AI.",
"id": "tulu3",
"name": "Tulu3",
"variants": [
{
"id": "tulu3:latest",
"parameters": "Unknown",
"size": "4.9GB"
},
{
"id": "tulu3:70b",
"parameters": "70B",
"size": "43GB"
},
{
"id": "tulu3:8b",
"parameters": "8B",
"size": "4.9GB"
}
],
"vision": false
},
{
"description": "A 3.8B model fine-tuned on a private high-quality synthetic dataset for information extraction, based on Phi-3.",
"id": "nuextract",
"name": "Nuextract",
"variants": [
{
"id": "nuextract:latest",
"parameters": "Unknown",
"size": "2.2GB"
},
{
"id": "nuextract:3.8b",
"parameters": "3.8B",
"size": "2.2GB"
}
],
"vision": false
},
{
"description": "MistralLite is a fine-tuned model based on Mistral with enhanced capabilities of processing long contexts.",
"id": "mistrallite",
"name": "Mistrallite",
"variants": [
{
"id": "mistrallite:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "mistrallite:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "A state-of-the-art fact-checking model developed by Bespoke Labs.",
"id": "bespoke-minicheck",
"name": "Bespoke Minicheck",
"variants": [
{
"id": "bespoke-minicheck:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "bespoke-minicheck:7b",
"parameters": "7B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "A top-performing mixture of experts model, fine-tuned with high-quality data.",
"id": "notux",
"name": "Notux",
"variants": [
{
"id": "notux:latest",
"parameters": "Unknown",
"size": "26GB"
},
{
"id": "notux:8x7b",
"parameters": "7B",
"size": "26GB"
}
],
"vision": false
},
{
"description": "A 7B chat model fine-tuned with high-quality data and based on Zephyr.",
"id": "notus",
"name": "Notus",
"variants": [
{
"id": "notus:latest",
"parameters": "Unknown",
"size": "4.1GB"
},
{
"id": "notus:7b",
"parameters": "7B",
"size": "4.1GB"
}
],
"vision": false
},
{
"description": "Wizard Vicuna is a 13B parameter model based on Llama 2 trained by MelodysDreamj.",
"id": "wizard-vicuna",
"name": "Wizard Vicuna",
"variants": [
{
"id": "wizard-vicuna:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "wizard-vicuna:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "Embedding model from BAAI mapping texts to vectors.",
"id": "bge-large",
"name": "Bge Large",
"variants": [
{
"id": "bge-large:latest",
"parameters": "Unknown",
"size": "671MB"
},
{
"id": "bge-large:335m",
"parameters": "335M",
"size": "671MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "An open weights function calling model based on Llama 3, competitive with GPT-4o function calling capabilities.",
"id": "firefunction-v2",
"name": "Firefunction V2",
"variants": [
{
"id": "firefunction-v2:latest",
"parameters": "Unknown",
"size": "40GB"
},
{
"id": "firefunction-v2:70b",
"parameters": "70B",
"size": "40GB"
}
],
"vision": false,
"tools": true
},
{
"description": "A new small LLaVA model fine-tuned from Phi 3 Mini.",
"id": "llava-phi3",
"name": "Llava Phi3",
"variants": [
{
"id": "llava-phi3:latest",
"parameters": "Unknown",
"size": "2.9GB"
},
{
"id": "llava-phi3:3.8b",
"parameters": "3.8B",
"size": "2.9GB"
}
],
"vision": true
},
{
"description": "A high-performing code instruct model created by merging two existing code models.",
"id": "codebooga",
"name": "Codebooga",
"variants": [
{
"id": "codebooga:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "codebooga:34b",
"parameters": "34B",
"size": "19GB"
}
],
"vision": false
},
{
"description": "Merge of the Open Orca OpenChat model and the Garage-bAInd Platypus 2 model. Designed for chat and code generation.",
"id": "open-orca-platypus2",
"name": "Open Orca Platypus2",
"variants": [
{
"id": "open-orca-platypus2:latest",
"parameters": "Unknown",
"size": "7.4GB"
},
{
"id": "open-orca-platypus2:13b",
"parameters": "13B",
"size": "7.4GB"
}
],
"vision": false
},
{
"description": "DBRX is an open, general-purpose LLM created by Databricks.",
"id": "dbrx",
"name": "Dbrx",
"variants": [
{
"id": "dbrx:latest",
"parameters": "Unknown",
"size": "74GB"
},
{
"id": "dbrx:132b",
"parameters": "132B",
"size": "74GB"
}
],
"vision": false
},
{
"description": "DeepSeek-OCR is a vision-language model that can perform token-efficient OCR.",
"id": "deepseek-ocr",
"name": "Deepseek Ocr",
"variants": [
{
"id": "deepseek-ocr:latest",
"parameters": "Unknown",
"size": "6.7GB"
},
{
"id": "deepseek-ocr:3b",
"parameters": "3B",
"size": "6.7GB"
}
],
"vision": true
},
{
"description": "A language model created by combining two fine-tuned Llama 2 70B models into one.",
"id": "goliath",
"name": "Goliath",
"variants": [
{
"id": "goliath:latest",
"parameters": "Unknown",
"size": "66GB"
},
{
"id": "goliath:120b-q2_k",
"parameters": "120B",
"size": "50GB"
},
{
"id": "goliath:120b-q3_k_s",
"parameters": "120B",
"size": "51GB"
},
{
"id": "goliath:120b-q3_k_m",
"parameters": "120B",
"size": "56GB"
},
{
"id": "goliath:120b-q3_k_l",
"parameters": "120B",
"size": "62GB"
},
{
"id": "goliath:120b-q4_0",
"parameters": "120B",
"size": "66GB"
},
{
"id": "goliath:120b-q4_1",
"parameters": "120B",
"size": "74GB"
},
{
"id": "goliath:120b-q4_k_s",
"parameters": "120B",
"size": "66GB"
},
{
"id": "goliath:120b-q4_k_m",
"parameters": "120B",
"size": "71GB"
},
{
"id": "goliath:120b-q5_0",
"parameters": "120B",
"size": "81GB"
},
{
"id": "goliath:120b-q5_1",
"parameters": "120B",
"size": "88GB"
},
{
"id": "goliath:120b-q5_k_s",
"parameters": "120B",
"size": "81GB"
},
{
"id": "goliath:120b-q5_k_m",
"parameters": "120B",
"size": "83GB"
},
{
"id": "goliath:120b-q6_k",
"parameters": "120B",
"size": "97GB"
},
{
"id": "goliath:120b-q8_0",
"parameters": "120B",
"size": "125GB"
},
{
"id": "goliath:120b-fp16",
"parameters": "120B",
"size": "236GB"
}
],
"vision": false
},
{
"description": "Nemotron 3 Nano - A new Standard for Efficient, Open, and Intelligent Agentic Models",
"id": "nemotron-3-nano",
"name": "Nemotron 3 Nano",
"variants": [
{
"id": "nemotron-3-nano:latest",
"parameters": "Unknown",
"size": "24GB"
},
{
"id": "nemotron-3-nano:30b",
"parameters": "30B",
"size": "24GB"
},
{
"id": "nemotron-3-nano:30b-cloud",
"parameters": "30B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Sailor2 are multilingual language models made for South-East Asia. Available in 1B, 8B, and 20B parameter sizes.",
"id": "sailor2",
"name": "Sailor2",
"variants": [
{
"id": "sailor2:latest",
"parameters": "Unknown",
"size": "5.2GB"
},
{
"id": "sailor2:20b",
"parameters": "20B",
"size": "12GB"
},
{
"id": "sailor2:8b",
"parameters": "8B",
"size": "5.2GB"
},
{
"id": "sailor2:1b",
"parameters": "1B",
"size": "1.1GB"
}
],
"vision": false
},
{
"description": "Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets.",
"id": "olmo-3",
"name": "Olmo 3",
"variants": [
{
"id": "olmo-3:latest",
"parameters": "Unknown",
"size": "4.5GB"
},
{
"id": "olmo-3:32b",
"parameters": "32B",
"size": "19GB"
},
{
"id": "olmo-3:7b",
"parameters": "7B",
"size": "4.5GB"
}
],
"vision": false
},
{
"description": "24B model that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
"id": "devstral-small-2",
"name": "Devstral Small 2",
"variants": [
{
"id": "devstral-small-2:latest",
"parameters": "Unknown",
"size": "15GB"
},
{
"id": "devstral-small-2:24b",
"parameters": "24B",
"size": "15GB"
},
{
"id": "devstral-small-2:24b-cloud",
"parameters": "24B",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"cloud": true
},
{
"description": "A new small reasoning model fine-tuned from the Qwen 2.5 3B Instruct model.",
"id": "smallthinker",
"name": "Smallthinker",
"variants": [
{
"id": "smallthinker:latest",
"parameters": "Unknown",
"size": "3.6GB"
},
{
"id": "smallthinker:3b",
"parameters": "3B",
"size": "3.6GB"
}
],
"vision": false
},
{
"description": "The smallest model in Cohere's R series delivers top-tier speed, efficiency, and quality to build powerful AI applications on commodity GPUs and edge devices.",
"id": "command-r7b",
"name": "Command R7b",
"variants": [
{
"id": "command-r7b:latest",
"parameters": "Unknown",
"size": "5.1GB"
},
{
"id": "command-r7b:7b",
"parameters": "7B",
"size": "5.1GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Phi 4 mini reasoning is a lightweight open model that balances efficiency with advanced reasoning ability.",
"id": "phi4-mini-reasoning",
"name": "Phi4 Mini Reasoning",
"variants": [
{
"id": "phi4-mini-reasoning:latest",
"parameters": "Unknown",
"size": "3.2GB"
},
{
"id": "phi4-mini-reasoning:3.8b",
"parameters": "3.8B",
"size": "3.2GB"
}
],
"vision": false
},
{
"description": "An upgraded version of DeekSeek-V2 that integrates the general and coding abilities of both DeepSeek-V2-Chat and DeepSeek-Coder-V2-Instruct.",
"id": "deepseek-v2.5",
"name": "Deepseek V2.5",
"variants": [
{
"id": "deepseek-v2.5:latest",
"parameters": "Unknown",
"size": "133GB"
},
{
"id": "deepseek-v2.5:236b",
"parameters": "236B",
"size": "133GB"
}
],
"vision": false
},
{
"description": "The IBM Granite Guardian 3.0 2B and 8B models are designed to detect risks in prompts and/or responses.",
"id": "granite3-guardian",
"name": "Granite3 Guardian",
"variants": [
{
"id": "granite3-guardian:latest",
"parameters": "Unknown",
"size": "2.7GB"
},
{
"id": "granite3-guardian:8b",
"parameters": "8B",
"size": "5.8GB"
},
{
"id": "granite3-guardian:2b",
"parameters": "2B",
"size": "2.7GB"
}
],
"vision": false
},
{
"description": "111 billion parameter model optimized for demanding enterprises that require fast, secure, and high-quality AI",
"id": "command-a",
"name": "Command A",
"variants": [
{
"id": "command-a:latest",
"parameters": "Unknown",
"size": "67GB"
},
{
"id": "command-a:111b",
"parameters": "111B",
"size": "67GB"
}
],
"vision": false,
"tools": true
},
{
"description": "An open large reasoning model for real-world solutions by the Alibaba International Digital Commerce Group (AIDC-AI).",
"id": "marco-o1",
"name": "Marco O1",
"variants": [
{
"id": "marco-o1:latest",
"parameters": "Unknown",
"size": "4.7GB"
},
{
"id": "marco-o1:7b",
"parameters": "7B",
"size": "4.7GB"
}
],
"vision": false
},
{
"description": "A robust conversational model designed to be used for both chat and instruct use cases.",
"id": "alfred",
"name": "Alfred",
"variants": [
{
"id": "alfred:latest",
"parameters": "Unknown",
"size": "24GB"
},
{
"id": "alfred:40b",
"parameters": "40B",
"size": "24GB"
}
],
"vision": false
},
{
"description": "Qwen3-Coder-Next is a coding-focused language model from Alibaba's Qwen team, optimized for agentic coding workflows and local development.",
"id": "qwen3-coder-next",
"name": "Qwen3 Coder Next",
"variants": [
{
"id": "qwen3-coder-next:latest",
"parameters": "Unknown",
"size": "52GB"
},
{
"id": "qwen3-coder-next:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
},
{
"id": "qwen3-coder-next:q4_k_m",
"parameters": "Unknown",
"size": "52GB"
},
{
"id": "qwen3-coder-next:q8_0",
"parameters": "Unknown",
"size": "85GB"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets.",
"id": "olmo-3.1",
"name": "Olmo 3.1",
"variants": [
{
"id": "olmo-3.1:latest",
"parameters": "Unknown",
"size": "19GB"
},
{
"id": "olmo-3.1:32b",
"parameters": "32B",
"size": "19GB"
}
],
"vision": false,
"tools": true
},
{
"description": "123B model that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
"id": "devstral-2",
"name": "Devstral 2",
"variants": [
{
"id": "devstral-2:latest",
"parameters": "Unknown",
"size": "75GB"
},
{
"id": "devstral-2:123b",
"parameters": "123B",
"size": "75GB"
},
{
"id": "devstral-2:123b-cloud",
"parameters": "123B",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "A new state-of-the-art version of the lightweight Command R7B model that excels in advanced Arabic language capabilities for enterprises in the Middle East and Northern Africa.",
"id": "command-r7b-arabic",
"name": "Command R7b Arabic",
"variants": [
{
"id": "command-r7b-arabic:latest",
"parameters": "Unknown",
"size": "5.1GB"
},
{
"id": "command-r7b-arabic:7b",
"parameters": "7B",
"size": "5.1GB"
}
],
"vision": false,
"tools": true
},
{
"description": "Kimi K2.5 is an open-source, native multimodal agentic model that seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as conversational and agentic paradigms.",
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"variants": [
{
"id": "kimi-k2.5:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "The Cogito v2.1 LLMs are instruction tuned generative models. All models are released under MIT license for commercial use.",
"id": "cogito-2.1",
"name": "Cogito 2.1",
"variants": [
{
"id": "cogito-2.1:latest",
"parameters": "Unknown",
"size": "1.3TB"
},
{
"id": "cogito-2.1:671b",
"parameters": "671B",
"size": "1.3TB"
},
{
"id": "cogito-2.1:671b-cloud",
"parameters": "671B",
"size": "Cloud (API only)"
}
],
"vision": false,
"cloud": true
},
{
"description": "gpt-oss-safeguard-20b and gpt-oss-safeguard-120b are safety reasoning models built-upon gpt-oss",
"id": "gpt-oss-safeguard",
"name": "Gpt Oss Safeguard",
"variants": [
{
"id": "gpt-oss-safeguard:latest",
"parameters": "Unknown",
"size": "14GB"
},
{
"id": "gpt-oss-safeguard:120b",
"parameters": "120B",
"size": "65GB"
},
{
"id": "gpt-oss-safeguard:20b",
"parameters": "20B",
"size": "14GB"
}
],
"vision": false,
"tools": true,
"thinking": true
},
{
"description": "FunctionGemma is a specialized version of Google's Gemma 3 270M model fine-tuned explicitly for function calling. ",
"id": "functiongemma",
"name": "Functiongemma",
"variants": [
{
"id": "functiongemma:latest",
"parameters": "Unknown",
"size": "301MB"
},
{
"id": "functiongemma:270m",
"parameters": "270M",
"size": "301MB"
}
],
"vision": false,
"tools": true
},
{
"description": "Advanced agentic, reasoning and coding capabilities.",
"id": "glm-4.6",
"name": "Glm 4.6",
"variants": [
{
"id": "glm-4.6:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Gemini 3 Flash offers frontier intelligence built for speed at a fraction of the cost.",
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"variants": [
{
"id": "gemini-3-flash-preview:latest",
"parameters": "Unknown",
"size": "Cloud (API only)"
},
{
"id": "gemini-3-flash-preview:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "MiniMax M2 is a high-efficiency large language model built for coding and agentic workflows.",
"id": "minimax-m2",
"name": "Minimax M2",
"variants": [
{
"id": "minimax-m2:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Advancing the Coding Capability",
"id": "glm-4.7",
"name": "Glm 4.7",
"variants": [
{
"id": "glm-4.7:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "A strong reasoning and agentic model from Z.ai with 744B total parameters (40B active), built for complex systems engineering and long-horizon tasks.",
"id": "glm-5",
"name": "Glm 5",
"variants": [
{
"id": "glm-5:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "nomic-embed-text-v2-moe is a multilingual MoE text embedding model that excels at multilingual retrieval.",
"id": "nomic-embed-text-v2-moe",
"name": "Nomic Embed Text V2 Moe",
"variants": [
{
"id": "nomic-embed-text-v2-moe:latest",
"parameters": "Unknown",
"size": "958MB"
}
],
"vision": false,
"model_type": "embedding"
},
{
"description": "GLM-OCR is a multimodal OCR model for complex document understanding, built on the GLM-V encoder–decoder architecture.",
"id": "glm-ocr",
"name": "Glm Ocr",
"variants": [
{
"id": "glm-ocr:latest",
"parameters": "Unknown",
"size": "2.2GB"
},
{
"id": "glm-ocr:q8_0",
"parameters": "Unknown",
"size": "1.6GB"
},
{
"id": "glm-ocr:bf16",
"parameters": "Unknown",
"size": "2.2GB"
}
],
"vision": true,
"tools": true
},
{
"description": "A state-of-the-art mixture-of-experts (MoE) language model. Kimi K2-Instruct-0905 demonstrates significant improvements in performance on public benchmarks and real-world coding agent tasks.",
"id": "kimi-k2",
"name": "Kimi K2",
"variants": [
{
"id": "kimi-k2:1t-cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "MiniMax-M2.5 is a state-of-the-art large language model designed for real-world productivity and coding tasks.",
"id": "minimax-m2.5",
"name": "Minimax M2.5",
"variants": [
{
"id": "minimax-m2.5:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. ",
"id": "deepseek-v3.2",
"name": "Deepseek V3.2",
"variants": [
{
"id": "deepseek-v3.2:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Kimi K2 Thinking, Moonshot AI's best open-source thinking model.",
"id": "kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"variants": [
{
"id": "kimi-k2-thinking:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"thinking": true,
"cloud": true
},
{
"description": "Exceptional multilingual capabilities to elevate code engineering",
"id": "minimax-m2.1",
"name": "Minimax M2.1",
"variants": [
{
"id": "minimax-m2.1:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": false,
"tools": true,
"cloud": true
},
{
"description": "A general-purpose multimodal mixture-of-experts model for production-grade tasks and enterprise workloads.",
"id": "mistral-large-3",
"name": "Mistral Large 3",
"variants": [
{
"id": "mistral-large-3:675b-cloud",
"parameters": "675B",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"cloud": true
},
{
"description": "The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency.",
"id": "qwen3.5",
"name": "Qwen3.5",
"variants": [
{
"id": "qwen3.5:397b-cloud",
"parameters": "397B",
"size": "Cloud (API only)"
},
{
"id": "qwen3.5:cloud",
"parameters": "Unknown",
"size": "Cloud (API only)"
}
],
"vision": true,
"tools": true,
"thinking": true,
"cloud": true
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment