Model Catalog
Pick a model from our selection, or deploy a previously fine-tuned model.
Laya
New Laya is a multilingual, non-autoregressive System One decision model for calibrated routing, scoring, guardrails, moderation, and other low-latency decisions.
DeepSeek V4.1 Flash
New DeepSeek V4.1 Flash is a Mixture-of-Experts model built for efficient reasoning, tool use, and agentic workloads.
DeepSeek V4 Flash Vision Exp
New DeepSeek V4 Flash Vision Exp is an experimental multimodal MoE model built for efficient visual reasoning, tool use, and agentic workloads. It matches the performance of DeepSeek V4 Flash while heavily increasing multimodal agent capabilities.
Qwen3.8 Flash Next FP8
New Qwen3.8-Flash-Next-FP8 is a native multimodal MoE model with 125B language-model parameters and 6B activated parameters, plus efficient n-gram embeddings and MTP for long-context agentic workloads.
GLM 5.3 Flash
New GLM 5.3 Flash is a native multimodal 320B-parameter MoE model with 18B activated parameters, combining sparse and linear attention for efficient long-context coding and agentic workloads.
Qwen3.8 27B
New Qwen3.8-27B is a compact 27B dense vision-language model with native image and video understanding, flexible thinking control, and stronger coding, research, and long-horizon agentic performance. Context length is 262K natively and extensible to 1M tokens.
NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4
New NVIDIA Nemotron 3.5 Lightning 30B A3B is a 30B-parameter sparse language model, provided in an NVIDIA-optimized NVFP4 format for efficient enterprise inference.
NVIDIA Nemotron 3.5 Lightning 30B A3B BF16
New NVIDIA Nemotron 3.5 Lightning 30B A3B is a 30B-parameter sparse language model, provided in BF16 format for enterprise inference.
Muse Glimmer 30B
New Muse Glimmer is a 30B-parameter causal language model, distilled from Muse Spark and built for autonomous local agents. It combines multimodal understanding, multi-step reasoning, reliable tool use, and failure recovery on consumer hardware.
DeepSeek V4 Flash 0731
New DeepSeek V4 Flash 0731 is an updated release of the 284B-parameter MoE model with 13B activated parameters, built for efficient long-context intelligence.
Gemma 4 31B IT FP8 Dynamic
RedHatAI's gemma-4-31B-it-FP8-dynamic was obtained by quantizing the weights and activations of google/gemma-4-31B-it to FP8 data type using dynamic per-token quantization.
Qwen AgentWorld 35B A3B
Qwen-AgentWorld-35B-A3B is a 35B parameter Qwen model with 3B active parameters, optimized for agentic workloads and long-context reasoning.
NVIDIA Nemotron 3 Ultra 550B A55B NVFP4
Nemotron 3 Ultra 550B A55B NVFP4 is an NVIDIA-optimized release for advanced reasoning, agentic tool use, and long-context conversational workloads.
MiniMax M2.7
MiniMax M2.7 is a 229B-parameter MoE model for software engineering, agentic tool use, productivity workflows, and long-context reasoning.
MiniMax M2.7 NVFP4
NVIDIA-optimized NVFP4 release of MiniMax M2.7, a 230B-parameter sparse MoE model for complex software engineering, agentic tool use, and productivity workflows.
Nemotron 3 Nano Omni 30B A3B Reasoning BF16
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 is a multimodal large language model that unifies video, audio, image, and text understanding to support enterprise-grade Q&A, summarization, transcription, and document intelligence workflows.
Nemotron 3 Nano Omni 30B A3B Reasoning FP8
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 is a multimodal large language model that unifies video, audio, image, and text understanding to support enterprise-grade Q&A, summarization, transcription, and document intelligence workflows.
Kimi K2.5 NVFP4
NVIDIA-optimized NVFP4 release of Moonshot AI Kimi K2.5: a large multimodal model (text, image, video) with up to 256k context, quantized with NVIDIA Model Optimizer for efficient inference.
Mistral Small 4 119B 2603
Mistral Small 4 is a frontier-class multimodal model in the Mistral Small line, with strong multilingual, vision, and long-context performance.
Mistral Small 4 119B 2603 NVFP4
NVFP4-quantized Mistral Small 4: a frontier-class multimodal model with strong multilingual, vision, and long-context performance at lower memory footprint.
Cohere Transcribe 03-2026
Cohere speech transcription model with great performance for automatic speech recognition and long-form audio.
Gemma 4 31B IT
Gemma 4 is Google's open multimodal family built on the same research stack as Gemini. This instruction-tuned 31B variant supports image-text input and conversational use.
Gemma 4 26B A4B IT
Gemma 4 26B A4B is a mixture-of-experts instruction-tuned model in the Gemma 4 line, with multimodal (image-text) conversational capabilities.
Gemma 4 E4B IT
Compact Gemma 4 E4B instruction-tuned variant with multimodal (image-text and broader modality) support for efficient deployment.
Gemma 4 E2B IT
Smallest Gemma 4 E2B instruction-tuned variant with multimodal (image-text and broader modality) support for edge-friendly deployment.
Qwen3.5 27B
Qwen3.5-27B is a mid-sized version of Qwen3.5 that retains its advanced multimodal capabilities, efficient architecture, strong RL-driven generalization, and broad multilingual support while offering a more balanced performance-efficiency tradeoff.
Qwen3.5 9B
Qwen3.5-9B is a smaller, efficient variant of Qwen3.5 that preserves its multimodal intelligence, scalable reasoning, and broad multilingual capabilities while optimizing for lower resource usage and faster deployment.
Qwen3 Coder Next
Qwen3-Coder-Next is an open-weight coding-focused model that delivers high performance with minimal active parameters, excels in agentic reasoning and tool use, and integrates seamlessly into real-world IDEs with long context support.
NVIDIA Nemotron 3 Super 120B A12B BF16
Nemotron-3-Super-120B-A12B-BF16 is a large language model (LLM) trained by NVIDIA, designed to deliver strong agentic, reasoning, and conversational capabilities.
NVIDIA Nemotron 3 Nano 30B A3B BF16
Nemotron-3-Nano-30B-A3B-BF16 is a large language model (LLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks.
Ministral 3 Small 14B Reasoning 2512
The largest model in the Ministral 3 family, Ministral 3 14B offers frontier capabilities and performance comparable to its larger Mistral Small 3.2 24B counterpart. A powerful and efficient language model with vision capabilities.
Mistral Large 3 675B Instruct 2512 NVFP4
Mistral Large 3 is a state-of-the-art general-purpose Multimodal granular Mixture-of-Experts model with 41B active parameters and 675B total parameters trained from the ground up with 3000 H200s.
GPT OSS 120B
GPT OSS is the latest OpenAI open-model designed for powerful reasoning, agentic tasks, and versatile developer use cases. GPT OSS 120B is intended for production, general purpose, and high reasoning use cases.
GPT OSS 20B
GPT OSS is the latest OpenAI open-model designed for powerful reasoning, agentic tasks, and versatile developer use cases. GPT OSS 20B is intended for lower latency, and local or specialized use cases.
DeepSeek V4 Flash
Deprecated DeepSeek V4 Flash is a 284B-parameter MoE model with 13B activated parameters and a one-million-token context window, built for efficient long-context intelligence.
IBM Granite 4.0 H Micro
Deprecated Granite-4.0-H-Micro is a 3B parameter long-context instruct model finetuned from Granite-4.0-H-Micro-Base. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
IBM Granite 4.0 H Tiny
Deprecated Granite-4.0-H-Tiny is a 7B parameter long-context instruct model finetuned from Granite-4.0-H-Tiny-Base. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
IBM Granite 4.0 H Small
Deprecated Granite-4.0-H-Small is a 32B parameter long-context instruct model finetuned from Granite-4.0-H-Small-Base. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Magistral Small 2507
Deprecated Building upon Mistral Small 3.1 (2503), with added reasoning capabilities, undergoing SFT from Magistral Medium traces and RL on top, it's a small, efficient reasoning model with 24B parameters
Grok 2.5
Deprecated Grok 2.5 is a model trained and used at xAI in 2024, recently released. Grok 2.5 was xAI's best model in 2024.
Llama 4 Scout 17B 16E Instruct
Deprecated The Llama 4 collection of models are natively multimodal AI models that enable text and multimodal experiences. These models leverage a mixture-of-experts architecture to offer industry-leading performance in text and image understanding.
gemma 3 27b it
Deprecated Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models.
gemma 3 12b it
Deprecated Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models.
gemma 3 4b it
Deprecated Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models.
gemma 3 1b it
Deprecated Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models.
deepseek r1 distill qwen 32b
Deprecated DeepSeek-R1 is an advanced reasoning system using reinforcement learning with cold-start initialization, achieving performance comparable to leading proprietary models. The release includes distilled variants like DeepSeek-R1-Distill-Qwen-32B, which outperforms OpenAI's GPT-4-mini across mathematical reasoning (92.3% MATH) and coding benchmarks while maintaining commercial usability. This 32B-parameter model sets new state-of-the-art results for dense models in its class through optimized knowledge distillation from the R1 framework.
deepseek r1 distill llama 70b
Deprecated DeepSeek-R1 is an advanced reasoning model using reinforcement learning with cold-start initialization, matching top-tier systems in math and coding performance. The release includes distilled variants like DeepSeek-R1-Distill-Llama-70B, which preserves 94.5% MATH accuracy in the efficient Llama-70B architecture while supporting commercial applications.
Llama 3.2 11B Vision Instruct
Deprecated The Llama-3.2-11B-Vision-Instruct is a multimodal large language model optimized for visual recognition, image reasoning, captioning, and answering questions about images, trained on 6 billion image-text pairs, with 11 billion parameters, and supported for commercial and research use in English.
Llama 3.2 90B Vision Instruct
Deprecated The Llama-3.2-90B-Vision-Instruct is a multimodal large language model optimized for visual recognition, image reasoning, captioning, and answering questions about images, trained on 6 billion image-text pairs, with 90 billion parameters, and supported for commercial and research use in English.
Llama 3.3 70B Instruct
Deprecated The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).
meta llama 3.1 70b instruct
Deprecated The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out)
Meta Llama 3.1 8b Instruct
Deprecated The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out)
Mixtral 8x22B Instruct v0.1
Deprecated The Mixtral-8x22B-Instruct-v0.1 LLM, a fine-tuned version of the high-efficiency, sparse Mixture-of-Experts Mixtral-8x22B model, excels in multilingual fluency, advanced math, coding skills, and scalable tech application development.
Meta Llama 3 8b Instruct
Deprecated Meta Llama 3 is a family of instruction-tuned, auto-regressive LLMs with optimized transformer architecture, offered in 8B and 70B sizes, designed for superior performance in dialogue applications and enhanced safety and helpfulness.
Meta Llama 3.1 405B Instruct FP8
Deprecated The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out)