Tested models
Every model we measure on business documents. Click a model to see its results, errors, and cost.
| Model | Vendor | Reads images | Open weights | List price per 1M tokens (input / output) |
|---|---|---|---|---|
| Claude Haiku 5.5 | Anthropic | yes | no | $0.10 / $0.50 |
| Claude Sonnet 5 | Anthropic | yes | no | $2.00 / $10.00 |
| Claude Sonnet 5.5 | Anthropic | yes | no | $2.00 / $10.00 |
| DeepSeek V4.1 Flash | DeepSeek | yes | yes | $0.30 / $1.20 |
| Gemini 3.5 Flash Lite | yes | no | $0.30 / $2.50 | |
| Gemini 3.8 Flash | yes | no | $0.75 / $3.75 | |
| Gemma 4 31B | yes | yes | $0.09 / $0.34 | |
| Mistral Large 4 | Mistral | yes | no | $0.68 / $2.09 |
| Mistral Medium 3.5 | Mistral | yes | no | $1.50 / $7.50 |
| Mistral Small 4 | Mistral | yes | yes | $0.15 / $0.60 |
| GPT-6 Luna | OpenAI | yes | no | $0.10 / $0.50 |
| GPT-6 Sol | OpenAI | yes | no | $2.00 / $10.00 |
| Qwen3.8 Flash | Qwen | yes | no | $0.15 / $0.47 |
| GLM-5.3 Flash | Z.ai | yes | yes | $0.15 / $0.50 |