Models

MiniMax M3 Preview is a multimodal MoE vision-language model with strong reasoning, coding, and tool-calling capabilities.

Diffusion-based 26B parameter LLM enabling parallel token generation for real-time text apps.

Open, efficient hybrid Mamba-Transformer MoE with 1M context, excelling in agentic reasoning, coding, planning, tool calling, and more.

Multilingual, multimodal model for detecting unsafe and toxic content.

Generates physics-aware videos from text prompts or an image prompt for physical AI development.

Vision language model that excels in understanding the physical world using structured reasoning on videos or images.

A sparse MoE multimodal reasoning model good for enterprise, agentic and coding tasks.

1T multimodal MoE for long-horizon coding, agentic tool use, and image/video understanding.

A high performing model for text generation, coding and agentic use cases.

Nemotron 3 Nano Omni is an omni-modal reasoning model that understands images, video, speech, text.

DeepSeek V4 Flash is a 284B MoE model with 1M-token context optimized for fast coding and agents.

DeepSeek V4 scales to 1M-token context windows with efficient MoE architecture for coding tasks.

GLM-5.1 is a flagship LLM for agentic workflows, coding, and long-horizon reasoning tasks.

Multilingual, multimodal model for detecting unsafe and toxic content.

NVIDIA Synthetic Video Detector is an AI-powered micro-service for detecting AI-generated (synthetic) videos.

Detect and track speaker identities across video frames.

Open VLM for quantum computer calibration chart understanding across a range of qubit modalities.

MiniMax M2.7 is a 230B-parameter text-to-text AI model excelling in coding, reasoning, and office tasks.

Dense 31B model delivering frontier reasoning for coding, agentic workflows, and fine-tuning.

Nemotron 3 Voicechat — real-time voice conversation AI.

Hybrid MoE model unifying instruct, reasoning, and coding with multimodal input and 256k context.

Open, efficient hybrid Mamba-Transformer MoE with 1M context, excelling in agentic reasoning, coding, planning, tool calling, and more.

122B MoE LLM (10B active) for coding, reasoning, multimodal chat. Agent-ready.

GLiNER PII detects Personally Identifiable Information in text.

Generates physics-aware video world states for physical AI development using text prompts and multiple spatial control inputs derived from real-world data or simulation.

Next-gen Qwen 3.5 VLM (400B MoE) brings advanced vision, chat, RAG, and agentic capabilities.

200B open-source reasoning engine with sparse MoE powering frontier agentic AI.

A context-aware safety model that applies reasoning to enforce domain-specific policies.

Open, efficient MoE model with 1M context, excelling in coding, reasoning, instruction following, tool calling, and more.

Translation model in 12 languages with few-shots example prompts capability.

A state-of-the-art general purpose MoE VLM ideal for chat, agentic and instruction based use cases.

A general purpose VLM ideal for chat and instruction based use cases.

StreamPETR offers efficient 3D object detection for autonomous driving by propagating sparse object queries temporally.

Leading multilingual content safety model for enhancing the safety and moderation capabilities of LLMs.

Nemotron Nano 12B v2 VL enables multi-image and video understanding, along with visual Q&A and summarization capabilities.

Japanese-specialized large-language-model for enterprises to read and understand complex business documents.

Qwen3-Next Instruct blends hybrid attention, sparse MoE, and stability boosts for ultra-long context AI.

ByteDance open-source LLM with long-context, reasoning, and agentic intelligence.

High-efficiency LLM with hybrid Transformer-Mamba design, excelling in reasoning and agentic tasks.

Mixture of Experts (MoE) reasoning LLM (text-only) designed to fit within 80GB GPU.

Smaller Mixture of Experts (MoE) text-only LLM for efficient AI reasoning and math.

High efficiency model with leading accuracy for reasoning, tool calling, chat, and instruction following.

Multilingual, hybrid-reasoning model optimized for Indian language tasks, programming, mathematical reasoning capabilities.

Generates high-quality numerical embeddings from text inputs.

High efficiency model with leading accuracy for reasoning, tool calling, chat, and instruction following.

End-to-end autonomous driving stack integrating perception, prediction, and planning with sparse scene representations for efficiency and safety.

Advanced transformer for multi-frame bird's-eye-view 3D perception in autonomous driving.

An MOE LLM that follows instructions, completes requests, and generates creative text.

A general purpose multimodal, multilingual 128 MoE model with 17B parameters.

An edge computing AI model which accepts text, audio and image input, ideal for resource-constrained environments.

An edge computing AI model which accepts text, audio and image input, ideal for resource-constrained environments.

Advanced state-of-the-art model with language understanding, superior reasoning, and text generation.

Multi-modal vision-language model that understands text/img and creates informative responses.

Generates physics-aware video world states for physical AI development using text prompts and multiple spatial control inputs derived from real-world data or simulation.

Multi-modal model to classify safety for input prompts as well output responses.

Leading reasoning and agentic AI accuracy model for PC and edge.

Removes unwanted noises from audio improving speech intelligibility.

Enhance input speech recorded with low-quality microphones in noisy or reverberant environments, producing studio-quality speech.

Expressive and engaging text-to-speech, generated from a short audio sample.

Advanced LLM for reasoning, math, general knowledge, and function calling.

Powers complex conversations with superior contextual understanding, reasoning and text generation.

Predicts the 3D structure of a protein from its amino acid sequence.

Built for agentic workflows, this model excels in coding, instruction following, and function calling.

Cutting-edge vision-language model exceling in high-quality reasoning from images.

Cutting-edge vision-language model exceling in high-quality reasoning from images.

The NV-EmbedCode model is a 7B Mistral-based embedding model optimized for code retrieval, supporting text, code, and hybrid queries.

Cutting-edge open multimodal model exceling in high-quality reasoning from image and audio inputs.

Advanced small language generative AI model for edge applications.

Advanced state-of-the-art small language model with language understanding, superior reasoning, and text generation.

Lightweight multilingual LLM powering AI applications in latency bound, memory/compute constrained environments.

Fine-tuned Llama 3.1 70B model for code generation, summarization, and multi-language tasks.

Advanced state-of-the-art small language model with language understanding, superior reasoning, and text generation.

Excels in NLP tasks, particularly in instruction-following, reasoning, and mathematics.

Generates embeddings of proteins from their amino acid sequences.

Most intelligent Claude model with extended thinking mode. Best for complex reasoning, coding, and agentic tasks.

Frontier model from xAI with real-time knowledge access. Strong at math, reasoning, and coding.

Reasoning-focused mini model. Faster and cheaper than Grok 3 while maintaining strong reasoning capability.

Online model with real-time web search. Returns cited answers with up-to-date information.

Fast, capable multimodal model with native tool use, spatial understanding, and real-time audio/video.

Reasoning model optimized for STEM and coding. Delivers high capability at a fraction of o1 pricing.

Reasoning variant of Gemini 2.0 Flash with visible thought process. Excels at complex multi-step problems.

Open-weight reasoning model rivalling o1 at a fraction of the cost. Excellent for math and coding.

GPU-accelerated model optimized for providing a probability score that a given passage contains the information to answer a question.

Large Mixture-of-Experts model with 456B parameters and 45.9B active. Massive 1M token context window.

671B parameter Mixture-of-Experts model with 37B active. Strong general-purpose performance at low cost.

Reasoning model from Alibaba with strong math and coding performance. Open-weight and cost-effective.

Frontier reasoning model that thinks before answering. Excels at complex math, science, and coding.

Versatile multilingual model with 72B parameters. Strong at code generation and multilingual tasks.

Code-specialized model matching GPT-4o on coding benchmarks. Supports 29+ programming languages.

Fast, affordable model with strong performance. Great for high-volume, latency-sensitive workloads.

Vision language model adept at comprehending text and visual inputs to produce informative responses.

Optimized SLM for on-device inference and fine-tuned for roleplay, RAG and function calling.

Fine-tune of Llama 3.1 405B with enhanced roleplay, creative writing, and agentic capabilities.

Affordable, intelligent small model for fast, lightweight tasks. Supports vision and function calling.

Multimodal flagship model with vision, audio, and text capabilities. Faster and cheaper than GPT-4 Turbo.

Llama 3 8B fine-tune trained on curated roleplay data. Excellent for creative and conversational use.

Previous-generation flagship with 128K context. Strong general-purpose model with vision support.

Powerful model for highly complex tasks. Top-tier performance on reasoning, analysis, and creative writing.

Balanced model for enterprise workloads. Good balance of speed and intelligence.

Mid-generation multimodal model with 2M token context. Strong at long-document and video understanding.

Fast and versatile model across modalities. Optimized for high-volume, cost-sensitive tasks.

Fast, inexpensive model for simple tasks. Good for classification, summarization, and basic chat.