Roboflow Vision Evals — OCR Transcription (October)
Multimodal
Character similarity for transcribed documents and text regions in the October OCR suite. Single preliminary run per effort tier; question count unreported.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 97.5 |
| Claude Opus 5.5 | 96.7 |
| GPT-6.1 Sol | 96.0 |
| Claude Fable 5.1 | 95.2 |
| Claude Haiku 5.5 | 94.7 |
| Gemini 3.8 Flash | 94.1 |
| Claude Opus 5 | 92.7 |
| GPT-5.6 Sol | 92.3 |
| Gemini 3.7 Flash | 91.9 |
| GPT-6 Sol | 91.0 |
| Gemini 3.1 Pro Preview | 90.8 |
| Gemini 3 Flash Preview | 90.5 |
| Gemini 3.6 Flash | 90.3 |
| Claude Fable 5 | 90.2 |
| Muse Spark 1.1 | 90.1 |
| Gemini 3.5 Flash | 90.0 |
| GPT-5.5 | 89.9 |
| Qwen3.7-Plus (source unspecified) | 89.7 |
| Grok 4.7 | 89.5 |
| Muse Spark 1.2 | 89.3 |
| Grok 4.6 | 89.0 |
| Gemini 3.5 Flash-Lite | 88.6 |
| Sonnet 5 | 88.6 |
| Gemini 2.5 Pro | 87.3 |
| Qwen3.8 Flash | 87.3 |
| Qwen3.5-27B | 86.1 |
| Grok 4.5 | 86.1 |
| Qwen3.7 Flash | 84.4 |
| Qwen3-VL-235B-A22B-Instruct | 84.3 |
| GPT-6 Luna | 84.0 |
| GPT-5.6 Luna | 83.5 |
| Muse Glimmer 30B | 82.9 |
| GLM-5.3 Flash | 81.5 |
| Kimi K3 | 80.4 |
| GPT-5.6 Terra | 80.2 |
| GPT-5.4 Mini | 78.5 |
| GLM 5V Turbo | 78.0 |
| Kimi K2.6 | 77.9 |
| Claude Sonnet 5.5 | 76.3 |
| Mistral Large 4 Preview | 75.3 |