# Labs

Browse foundation models by research lab, and see how widely they are offered.

Append .md to this page’s address to get the Markdown text for programs and AI.

- `/en/labs.md`
- `/labs.md`

67 items

| Lab | ID | Description | Models | Providers |
| --- | --- | --- | --- | --- |
| OpenAI | openai | The GPT family covers reasoning, code, multimodal, and agent applications. | 69 | 46 |
| Alibaba | alibaba | The Qwen family covers reasoning, code, vision, audio, and agent workflows. | 59 | 5 |
| Google | google | Gemini and Gemma span frontier multimodal reasoning and open-weight options. | 56 | 32 |
| Mistral | mistral | Balances open research and enterprise deployment across chat, code, and document intelligence. | 49 | 7 |
| Anthropic | anthropic | The Claude family emphasizes reliability, interpretability, and long-horizon agent tasks. | 28 | 21 |
| NVIDIA | nvidia | The Nemotron family offers open weights, training recipes, and accelerated deployment. | 23 | 11 |
| DeepSeek | deepseek | Known for cost-efficient reasoning, visible chain-of-thought, and coding ability. | 21 | 30 |
| Cohere | cohere | The Command family targets enterprise retrieval, multilingual use, and office agents. | 20 | 8 |
| xAI | xai | The Grok family emphasizes tool use, low-hallucination reasoning, code, and media interfaces. | 19 | 10 |
| Zhipu AI | zhipuai | The GLM family emphasizes long-horizon coding, terminal tasks, and cost-efficient hybrid reasoning. | 19 | 0 |
| Meta | meta | Llama and Muse emphasize open weights, multilingual support, and customizable deployment. | 17 | 14 |
| ByteDance Seed | bytedance-seed | General-purpose and character models from ByteDance’s Seed lab. | 14 | 4 |
| MiniMax | minimax | Agent models for coding, office work, and multimodal productivity. | 13 | 2 |
| Moonshot AI | moonshotai | The Kimi family excels at long-context agents, multimodal coding, and high-throughput development. | 11 | 28 |
| Amazon | amazon | No description yet | 9 | 5 |
| Xiaomi | xiaomi | The MiMo family targets coding agents, long context, and multimodal interaction. | 9 | 11 |
| Tencent | tencent | Hunyuan and Hy families target reasoning, code, long context, and agents. | 6 | 12 |
| DeepReinforce | deepreinforce | The Ornith family of self-scaffolding models for coding agents. | 5 | 0 |
| Inclusionai | inclusionai | No description yet | 5 | 5 |
| Sakana AI | sakana | A research lab known for model routing and multi-agent systems. | 5 | 10 |
| Aion Labs | aion-labs | No description yet | 4 | 4 |
| Arcee AI | arcee-ai | Open-weight models for efficient reasoning, tool use, and deployable intelligence. | 4 | 4 |
| Microsoft | microsoft | Microsoft’s model families for coding and productivity. | 4 | 4 |
| Nousresearch | nousresearch | No description yet | 4 | 4 |
| Perplexity | perplexity | Sonar treats search as a first-class capability, emphasizing cited, real-time answers. | 4 | 8 |
| Poolside | poolside | Open foundation models and a continuous-improvement system for software engineering. | 4 | 8 |
| StepFun | stepfun | The Step family targets fast multimodal agents and tool orchestration. | 4 | 9 |
| Baidu | baidu | No description yet | 3 | 2 |
| IBM | ibm | Granite and related models for enterprise deployment and governance. | 3 | 1 |
| Sao10k | sao10k | No description yet | 3 | 3 |
| Thedrummer | thedrummer | No description yet | 3 | 3 |
| Upstage | upstage | The Solar family targets document intelligence and efficient general chat. | 3 | 5 |
| Vispark | vispark | No description yet | 3 | 1 |
| Ai21 | ai21 | No description yet | 2 | 0 |
| Baai | baai | No description yet | 2 | 2 |
| Inception | inception | No description yet | 2 | 4 |
| Inference Net | inference-net | No description yet | 2 | 5 |
| Meituan | meituan | Meituan’s research on long-context and agent-oriented models. | 2 | 3 |
| Mindlab | mindlab | No description yet | 2 | 0 |
| Morph | morph | No description yet | 2 | 4 |
| Openbmb | openbmb | No description yet | 2 | 0 |
| Quiverai | quiverai | No description yet | 2 | 1 |
| Rekaai | rekaai | No description yet | 2 | 3 |
| Relace | relace | No description yet | 2 | 3 |
| Sarvam AI | sarvam | Open reasoning models for Indian languages and local enterprise use. | 2 | 1 |
| Swiss AI | swiss-ai | Switzerland’s open-model initiative, emphasizing transparency and reproducible research. | 2 | 0 |
| Thinking Machines | thinkingmachines | Inkling and related research models for ultra-long context and open weights. | 2 | 13 |
| Writer | writer | No description yet | 2 | 3 |
| AI Singapore | aisingapore | An open-source lab focused on Southeast Asian languages and regional use cases. | 1 | 0 |
| Anthracite Org | anthracite-org | No description yet | 1 | 4 |
| Apodex | apodex | No description yet | 1 | 1 |
| Baichuan | baichuan | No description yet | 1 | 1 |
| Cognitivecomputations | cognitivecomputations | No description yet | 1 | 3 |
| Fireworks | fireworks | No description yet | 1 | 5 |
| Gryphe | gryphe | No description yet | 1 | 4 |
| Mancer | mancer | No description yet | 1 | 3 |
| Mixedbread | mixedbread | No description yet | 1 | 1 |
| Motif Technologies | motif-technologies | No description yet | 1 | 1 |
| Nex Agi | nex-agi | No description yet | 1 | 0 |
| Paddlepaddle | paddlepaddle | No description yet | 1 | 1 |
| Perceptron | perceptron | No description yet | 1 | 3 |
| Reflection | reflection | No description yet | 1 | 2 |
| SDAIA | sdaia | Language models associated with the Saudi Data and AI Authority. | 1 | 0 |
| Trendyol | trendyol | A lab working on e-commerce and local-language scenarios. | 1 | 0 |
| Unbiased | unbiased | No description yet | 1 | 2 |
| Undi95 | undi95 | No description yet | 1 | 4 |
| Vivgrid | vivgrid | No description yet | 1 | 0 |