AI Models by Provider

OpenAI is the San Francisco lab behind the GPT series and ChatGPT. Its frontier models are closed source and accessed through a paid API, while its open-weight Whisper speech models and the open GPT-OSS series ship under permissive licenses.

Alibaba publishes the Qwen series of open-weight models in dense and Mixture-of-Experts variants. Qwen leads many open leaderboards for code, math, and multilingual benchmarks.

Google ships closed-source Gemini models through Vertex AI and the Gemini app, plus open-weight Gemma models for self-hosting. The lineup spans text, vision, video, and on-device variants.

Anthropic builds Claude, a family of safety-focused frontier models offered through API and the Claude.ai product. All Claude models are closed source and priced per token.

xAI is Elon Musk’s AI lab building the Grok model family. Grok models are closed source and accessed through the X platform and an API.

Meta released the Llama family of open-weight models, which became the most widely fine-tuned foundation models in the world. Llama models ship under a community license that permits commercial use at most scales.

NVIDIA publishes the Nemotron, Cosmos, and Parakeet model families for language, world simulation, and speech recognition. Most NVIDIA-trained models are open weight.

Mistral AI is a Paris-based lab releasing both open-weight models (Mistral, Mixtral, Codestral, Devstral) and closed flagship models. Their MoE architectures pioneered cost-effective open inference.

Tencent releases the Hunyuan model family covering large language models, image generation, and video generation. Several Hunyuan models are open weight.

Z.ai is the brand of Zhipu AI, the Beijing lab behind the GLM open-weight model family and the CogVideoX / CogView visual generators. Z.ai serves both open weights and a closed API tier.

Black Forest Labs builds the FLUX family of state-of-the-art text-to-image models. They release both open-weight variants for self-hosting and closed-source pro tiers via API.

DeepSeek is a Chinese research lab known for cost-efficient Mixture-of-Experts models. Its open-weight V3 and R1 reasoning models match or exceed many closed frontier models on coding and math.

CodeFuse-AI (Ant Group) is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Moonshot AI is the Beijing lab behind the Kimi family of long-context, agentic models. Recent Kimi K2 releases ship under permissive open weights and lead many open agent leaderboards.

Amazon is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

MiniMax is a Shanghai-based lab shipping the MiniMax-M1 / MiniMax-Text language family and the Hailuo text-to-video generator. Several MiniMax models are open weight; the consumer products run through their app.

Microsoft Research publishes the Phi family of small open-weight models that punch above their parameter count on reasoning benchmarks. Phi models are designed for edge and on-device inference.

ByteDance, the TikTok parent, publishes the Doubao and Seedream model families spanning text, image, and video. Some Seedream and ASR models are open weight.

Xiaomi is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

BidirLM is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Qwen/Alibaba is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Hugging Face is the AI research hub and platform behind Parler-TTS, SmolLM, ZeroGPU, and the largest open model registry on the web.

IBM publishes the Granite family of open-weight language, code, and speech models under permissive licenses, plus enterprise embedding and reranker variants tuned for retrieval-augmented generation.

MyShell AI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Stepfun (StepFun) is a Shanghai AI lab shipping the Step language model series and the Step1X-Image / Step-Video generators. Several Step models are released under open weights.

Baidu is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Midjourney is a closed-source text-to-image model accessed through their Discord and web app. Midjourney v6 and v7 are among the highest-scoring image generators on aesthetic benchmarks.

Cohere builds enterprise-focused models including the Command family for chat and Embed for retrieval. Most Cohere models are closed source with a hosted API.

ElevenLabs builds closed-source text-to-speech and voice cloning models. Their multilingual voice models are the most widely deployed commercial TTS APIs.

intfloat (Microsoft Research) is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

GritLM (Contextual AI) is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Jina AI publishes open-weight multimodal embedding and reranker models, including Jina Embeddings v2 / v3 and ColBERT-v2 retrievers used widely in production RAG stacks.

Fish Audio publishes Fish Speech, an open-weight low-latency multilingual text-to-speech family with voice cloning, popular for self-hosted real-time speech.

Hexgrad publishes Kokoro, an Apache-licensed text-to-speech model family that achieves high voice quality at sub-100M parameter counts, popular for on-device TTS.

Kandinsky is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

The Technology Innovation Institute in Abu Dhabi publishes the Falcon family of open-weight foundation models, including dense, MoE, and small on-device variants under permissive licenses.

Voyage AI builds closed-source embedding and reranker models tuned for retrieval-augmented generation, with strong scores on MTEB and domain-specific retrieval benchmarks.

Kuaishou is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Runway is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Salesforce is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

ICT-CAS TIME / Querit is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Linq AI Research is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Octen AI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

ESPnet is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Efficient Speech is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Nyra Health is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Useful Sensors publishes Moonshine, an open-weight low-latency speech-to-text family designed for on-device transcription on phones, laptops, and microcontrollers.

Kyutai is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

The University of Texas at Austin is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

RVC-Boss is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Coqui shipped the XTTS family of open-weight multilingual voice cloning TTS models, widely self-hosted for high-quality speech synthesis in dozens of languages.

Maya Research is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

PolyAI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

ShoukanLabs is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Collabora is the open-source consultancy behind WhisperSpeech, an open-weight text-to-speech model trained with the Whisper backbone for fully open voice generation.

Columbia University is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

MetaVoice is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Genmo AI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Lightricks publishes the LTX-Video family of fast open-weight text-to-video and image-to-video diffusion transformers tuned for real-time generation on consumer GPUs.

Stability AI is the company behind Stable Diffusion, the open-weight text-to-image model that kicked off the generative imagery wave. SD3, SDXL, and Stable Video Diffusion are widely self-hosted.

kai-os is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

01.AI is the Beijing lab founded by Kai-Fu Lee, behind the Yi family of bilingual open-weight language models. Yi releases compete at the open frontier on Chinese and English benchmarks.

Inception AI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Microsoft AI is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.

Meituan is the AI lab behind the models on this page. Browse their full catalog ranked by benchmarks, popularity, and hardware fit.