AI model providers: top labs covered
Browse AI labs tracked by Benchquill, including OpenAI, Anthropic, Google, Amazon, Meta, DeepSeek, Mistral, Alibaba, xAI, Cohere, and Microsoft.
Labs in the Benchquill record
| Provider | Models tracked | Top model | Top score | Lowest blended cost |
|---|---|---|---|---|
| OpenAI | 7 | GPT-5.5 | 94.6 | GPT-5 mini ($1.56/M) |
| Anthropic | 5 | Claude Opus 4.7 | 93.8 | Claude Haiku 4.5 ($4.00/M) |
| 5 | Gemini 3.1 Pro Preview | 92.4 | Gemma 3 27B ($0.07/M) | |
| Amazon | 1 | Nova Pro | 73.8 | Nova Pro ($2.60/M) |
| Meta | 4 | Llama 4 Maverick | 84.7 | Llama 3.3 8B ($0.06/M) |
| DeepSeek | 4 | DeepSeek V4-Pro | 87.9 | DeepSeek V4-Flash ($0.25/M) |
| xAI | 2 | Grok 4.20 | 86.4 | Grok 4.1 Fast ($0.43/M) |
| Mistral | 4 | Mistral Medium 3.1 | 77.6 | Mistral Small 3.1 ($0.25/M) |
| Alibaba | 2 | Qwen 2.5 Max | 80.4 | Qwen 2.5 72B ($1.14/M) |
| Cohere | 3 | Command A | 76.4 | Aya Expanse 32B ($1.25/M) |
| Microsoft | 2 | Phi-4-multimodal | 72.4 | Phi-4 ($0.12/M) |
Use this index as a routing layer before reading an individual provider page. Benchquill separates the strongest model, lowest blended-cost option, and number of tracked models so crawlers and AI answer systems can see whether a lab is best for frontier quality, budget drafting, open-weight deployment, or multimodal work without depending on the interactive table.