# ParseBench leaderboard

Multimodal & Grounded benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/multimodal-grounded-parse-bench-llama-index

ParseBench: document parsing for AI agents (LlamaIndex leaderboard). Every model the catalog carries a published ParseBench value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Multimodal & Grounded |
| Measure | Deterministic test rules per page; five dimension scores and their mean, in percent |
| Tasks | 2,078 human-verified enterprise pages from insurance, finance and government documents |
| Difficulty | No method is strong on all five dimensions; charts and visual grounding separate the field |
| Published by | [ParseBench: A Document Parsing Benchmark for AI Agents](https://arxiv.org/abs/2604.08538) |

Scores document parsers and vision-language models on 2,078 human-verified enterprise pages across tables, charts, content faithfulness, semantic formatting and visual grounding, with the listed cost per page.

## ParseBench ranking

38 models with a published ParseBench value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Deterministic test rules per page; five dimension scores and their mean, in percent | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Claude Opus 5.5](https://modelscale.dev/models/claude-opus-5-5) | Anthropic | 79.85 | Observed 2026-10-23 · source benchlm:models |
| 2 | [Claude Fable 5.1](https://modelscale.dev/models/claude-fable-5-1) | Anthropic | 78.92 | Observed 2026-10-23 · source benchlm:models |
| 3 | [GPT-5.6 Sol](https://modelscale.dev/models/gpt-5-6-sol) | OpenAI | 75.35 | Observed 2026-10-23 · source benchlm:models |
| 4 | [Gemini 3 Flash](https://modelscale.dev/models/gemini-3-flash) | Google | 75.05 | Observed 2026-10-23 · source benchlm:models |
| 5 | [Gemini 3.5 Flash](https://modelscale.dev/models/gemini-3-5-flash) | Google | 74.02 | Observed 2026-10-23 · source benchlm:models |
| 6 | [Gemini 3.8 Flash](https://modelscale.dev/models/gemini-3-8-flash) | Google | 72.08 | Observed 2026-10-23 · source benchlm:models |
| 7 | [Gemini 3.7 Flash](https://modelscale.dev/models/gemini-3-7-flash) | Google | 71.27 | Observed 2026-10-23 · source benchlm:models |
| 8 | [Qwen3.8-27B](https://modelscale.dev/models/qwen3-8-27b) | Alibaba | 70.79 | Observed 2026-10-23 · source benchlm:models |
| 9 | [Claude Fable 5](https://modelscale.dev/models/claude-fable-5) | Anthropic | 70.78 | Observed 2026-10-23 · source benchlm:models |
| 10 | [GLM-5.3-Flash](https://modelscale.dev/models/glm-5-3-flash) | Z.AI | 70.75 | Observed 2026-10-23 · source benchlm:models |
| 11 | [GPT-6 Astra](https://modelscale.dev/models/gpt-6-astra) | OpenAI | 70.67 | Observed 2026-10-23 · source benchlm:models |
| 12 | [GPT-6.1 Sol](https://modelscale.dev/models/gpt-6-1-sol) | OpenAI | 70.48 | Observed 2026-10-23 · source benchlm:models |
| 13 | [Claude Sonnet 5.5](https://modelscale.dev/models/claude-sonnet-5-5) | Anthropic | 70.18 | Observed 2026-10-23 · source benchlm:models |
| 14 | [Gemini 3.6 Flash](https://modelscale.dev/models/gemini-3-6-flash) | Google | 70.03 | Observed 2026-10-23 · source benchlm:models |
| 15 | [GPT-6 Sol](https://modelscale.dev/models/gpt-6-sol) | OpenAI | 69.69 | Observed 2026-10-23 · source benchlm:models |
| 16 | [Gemini 3.1 Pro](https://modelscale.dev/models/gemini-3-1-pro) | Google | 69.14 | Observed 2026-10-23 · source benchlm:models |
| 17 | [GPT-5.6 Terra](https://modelscale.dev/models/gpt-5-6-terra) | OpenAI | 68.39 | Observed 2026-10-23 · source benchlm:models |
| 18 | [GPT-5.6 Luna](https://modelscale.dev/models/gpt-5-6-luna) | OpenAI | 68.34 | Observed 2026-10-23 · source benchlm:models |
| 19 | [GPT-5.5](https://modelscale.dev/models/gpt-5-5) | OpenAI | 67.76 | Observed 2026-10-23 · source benchlm:models |
| 20 | [GPT-6 Luna](https://modelscale.dev/models/gpt-6-luna) | OpenAI | 65.77 | Observed 2026-10-23 · source benchlm:models |
| 21 | [DeepSeek V4.1 Flash 1017](https://modelscale.dev/models/deepseek-v4-1-flash) | DeepSeek | 65.72 | Observed 2026-10-23 · source benchlm:models |
| 22 | [Claude Opus 4.8](https://modelscale.dev/models/claude-opus-4-8) | Anthropic | 63.70 | Observed 2026-10-23 · source benchlm:models |
| 23 | [Claude Opus 4.7](https://modelscale.dev/models/claude-opus-4-7) | Anthropic | 63.34 | Observed 2026-10-23 · source benchlm:models |
| 24 | [Gemma 4 31B](https://modelscale.dev/models/gemma-4-31b) | Google | 62.40 | Observed 2026-10-23 · source benchlm:models |
| 25 | [GPT-5.4](https://modelscale.dev/models/gpt-5-4) | OpenAI | 62.23 | Observed 2026-10-23 · source benchlm:models |
| 26 | [Claude Sonnet 5](https://modelscale.dev/models/claude-sonnet-5) | Anthropic | 62.13 | Observed 2026-10-23 · source benchlm:models |
| 27 | [Gemini 3.5 Flash-Lite](https://modelscale.dev/models/gemini-3-5-flash-lite) | Google | 60.64 | Observed 2026-10-23 · source benchlm:models |
| 28 | [Gemma 4 26B A4B](https://modelscale.dev/models/gemma-4-26b-a4b) | Google | 58.50 | Observed 2026-10-23 · source benchlm:models |
| 29 | [Gemini 3.1 Flash-Lite](https://modelscale.dev/models/gemini-3-1-flash-lite) | Google | 58.32 | Observed 2026-10-23 · source benchlm:models |
| 30 | [Claude Haiku 5.5](https://modelscale.dev/models/claude-haiku-5-5) | Anthropic | 57.40 | Observed 2026-10-23 · source benchlm:models |
| 31 | [Claude Opus 4.6](https://modelscale.dev/models/claude-opus-4-6) | Anthropic | 54.07 | Observed 2026-10-23 · source benchlm:models |
| 32 | [Claude Haiku 4.5 Thinking](https://modelscale.dev/models/claude-haiku-4-5-thinking) | Anthropic | 53.12 | Observed 2026-10-23 · source benchlm:models |
| 33 | [Nemotron 3 Nano Omni 30B A3B](https://modelscale.dev/models/nemotron-3-nano-omni-30b-a3b) | NVIDIA | 48.51 | Observed 2026-10-23 · source benchlm:models |
| 34 | [Claude Haiku 4.5](https://modelscale.dev/models/claude-haiku-4-5) | Anthropic | 45.17 | Observed 2026-10-23 · source benchlm:models |
| 35 | [Qwen3.6-35B-A3B](https://modelscale.dev/models/qwen3-6-35b-a3b) | Alibaba | 44.10 | Observed 2026-10-23 · source benchlm:models |
| 36 | [GPT-5.4 nano](https://modelscale.dev/models/gpt-5-4-nano) | OpenAI | 43.35 | Observed 2026-10-23 · source benchlm:models |
| 37 | [Gemma 4 E4B](https://modelscale.dev/models/gemma-4-e4b) | Google | 40.50 | Observed 2026-10-23 · source benchlm:models |
| 38 | [Amazon Nova 2 Lite](https://modelscale.dev/models/nova-2-lite) | Amazon | 40.43 | Observed 2026-10-23 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.