Benchmark
OmniDocBench 1.5
What is OmniDocBench 1.5?
OmniDocBench 1.5 is a comprehensive benchmark for evaluating multimodal large language models on document understanding tasks, including OCR, document parsing…
- Modality
- multimodal
- Evaluates
- multimodal, reasoning, structured output, vision
- Openness
- unknown
- Importer
- LLM Stats
- Reported scores
- 18
Source provenance
OmniDocBench 1.5 results and reported scores
LLM Stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DiffusionGemma 26B-A4B | 0.319 | 2026-06-10 | evidence | |
| Gemini 3 Flash | 0.121 | 2025-12-17 | evidence | |
| Gemini 3 Pro | 0.115 | 2025-11-18 | evidence | |
| Gemma 4 12B | 0.164 | 2026-05-23 | evidence | |
| GPT-5.4 | OpenAI | 0.891 | 2026-03-05 | evidence |
| GPT-5.4 mini | OpenAI | 0.8737 | 2026-03-17 | evidence |
| GPT-5.4 nano | OpenAI | 0.7581 | 2026-03-17 | evidence |
| GPT-5.5 Instant | OpenAI | 0.875 | 2026-05-05 | evidence |
| Kimi K2.5 | Moonshot AI | 0.888 | 2026-01-27 | evidence |
| MiniMax M3 | MiniMax | 0.916 | 2026-06-01 | evidence |
| Muse Glimmer-30B | Meta | 0.758 | 2026-08-10 | evidence |
| Qwen3.5-122B-A10B | Qwen | 0.898 | 2026-02-24 | evidence |
| Qwen3.5-27B | Qwen | 0.889 | 2026-02-24 | evidence |
| Qwen3.5-35B-A3B | Qwen | 0.893 | 2026-02-24 | evidence |
| Qwen3.6 Plus | Qwen | 0.912 | 2026-04-02 | evidence |
| Qwen3.6-35B-A3B | Qwen | 0.899 | 2026-04-16 | evidence |
| Qwen3.7-Plus | Qwen | 0.914 | 2026-05-31 | evidence |
| Qwen3.8-27B | Qwen | 0.911 | 2026-08-14 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.