Capability index
16 models across 9 providers. Seed scores today; replaced by real production eval results as the data moat grows.
| Model | Provider | Avg capability | Best for | Context | Speed | Cost / call* |
|---|---|---|---|---|---|---|
| Claude Fable 5 | Anthropic | 96 | Code / review, Hard reasoning / analysis | 300K | 40 | $0.040 |
| Claude Opus 4.8 | Anthropic | 95 | Code / review, Writing / marketing | 200K | 45 | $0.020 |
| GPT-5.5 | OpenAI | 94 | Hard reasoning / analysis, Summarize | 400K | 60 | $0.007 |
| Kimi K3open | Moonshot | 94 | Research / long context, Code / review | 1000K | 45 | $0.012 |
| Claude Sonnet 5 | Anthropic | 93 | Code / review, Summarize | 200K | 90 | $0.012 |
| GPT-5 | OpenAI | 93 | Hard reasoning / analysis, Classify / route | 400K | 60 | $0.007 |
| Claude Sonnet 4.6 | Anthropic | 92 | Code / review, Summarize | 200K | 90 | $0.012 |
| Gemini 2.5 Pro | 90 | Research / long context, Summarize | 1000K | 70 | $0.007 | |
| Grok 4 | xAI | 89 | Hard reasoning / analysis, Classify / route | 256K | 80 | $0.012 |
| Gemini 2.5 Flash | 85 | Research / long context, Classify / route | 1000K | 180 | $0.002 | |
| GPT-5 mini | OpenAI | 85 | Classify / route, Extract structured data | 400K | 120 | $0.001 |
| Claude Haiku 4.5 | Anthropic | 85 | Classify / route, Extract structured data | 200K | 160 | $0.004 |
| DeepSeek V3open | DeepSeek | 85 | Classify / route, Code / review | 128K | 90 | $0.001 |
| Qwen 3 Maxopen | Alibaba | 85 | Classify / route, Extract structured data | 256K | 120 | $0.001 |
| Mistral Large | Mistral | 84 | Classify / route, Extract structured data | 128K | 100 | $0.006 |
| Llama 4 Maverickopen | Meta | 82 | Classify / route, Extract structured data | 256K | 130 | $0.001 |
*Representative 1,500-in / 500-out call. Prices illustrative — refresh from provider price lists.