Vision models, ranked
Every model here accepts image input. If your documents are in Persian, resolution decides the outcome more than the model does: send a high-DPI scan and skip aggressive JPEG compression. Pair it with a model that also supports structured output and the fields come back as JSON instead of prose.
Most used in this category
- 1GPT-4o-mini118 tok174,075
Best value in this category
Score out of 100: 50% cheapness, 30% capabilities, 20% context size — not a quality benchmark.
| # | Model | Context | ≈ 1,000 words | Score |
|---|---|---|---|---|
| 1 | Gemini 2.5 Flash Lite (batch) | 1,048,576 | 94 | 86.9 |
| 2 | Qwen3.7 Flash | 1,000,000 | 60 | 85.6 |
| 3 | Muse Spark 1.2 Contributor | 1,048,576 | 113 | 84.9 |
| 4 | Muse Spark 1.3 Contributor | 1,048,576 | 113 | 84.9 |
| 5 | GPT-5 Nano (batch) | 400,000 | 85 | 84.7 |
| 6 | Nex-N2-Mini | 262,144 | 47 | 83.6 |
| 7 | GPT-4.1 Nano (batch) | 1,047,576 | 94 | 80.9 |
| 8 | GLM Flash Latest | 1,310,720 | 116 | 79.4 |
| 9 | Gemini 2.5 Flash Lite | 1,048,576 | 189 | 79.4 |
| 10 | GLM 5.3 Flash | 1,310,720 | 123 | 78.8 |
| 11 | Qwen3.5-Flash | 1,000,000 | 123 | 77.9 |
| 12 | GPT-5 Nano | 400,000 | 170 | 77.2 |
| 13 | Qwen3.5-9B | 262,144 | 94 | 76.1 |
| 14 | GPT-5.6 Luna Pro (batch) | 1,050,000 | 264 | 75.7 |
| 15 | GPT-5.6 Luna (batch) | 1,050,000 | 264 | 75.7 |