AmbiHTMLBench.
A multidimensional benchmark for image-to-HTML reconstruction of heterogeneous documents.
Leaderboard
Objective
16 modelsObj. score ↑
| Rank | Model | Score ↑ |
|---|---|---|
| 1 |  Gemini 3.1 Pro Preview |
80.92 |
| 2 |  Claude Opus 5.5 |
80.48 |
| 3 | Gemini 3.8 Flash |
78.88 |
| 4 |  Kimi-K3 |
77.32 |
| 5 | Claude Opus 5 |
77.12 |
| 6 |  Qwen3.8-Max |
76.33 |
| 7 |  GPT-6.1-Sol |
75.89 |
| 8 | GPT-6-Astra |
75.55 |
| 9 | Qwen3.7-Plus |
75.05 |
| 10 | Claude Sonnet 5.5 |
75.00 |
| 11 | GPT-5.6-Sol |
74.61 |
| 12 | GPT-5.6-Terra |
74.45 |
| 13 | Claude Sonnet 5 |
74.33 |
| 14 |  ERNIE-5.0 |
74.10 |
| 15 |  GLM-5V-Turbo |
73.30 |
| 16 |  Grok-4.5 |
35.04 |
Subjective
10 modelsBT score ↑
| Rank | Model | BT score ↑(95% CI) | Soft preferencerate | Rankinterval |
|---|---|---|---|---|
| 1 | GPT-5.6-Sol |
154.70[140.59, 168.82] | 71.64% | [1, 1] |
| 2 | GPT-5.6-Terra |
102.87[90.18, 116.25] | 64.76% | [2, 3] |
| 3 | Gemini 3.1 Pro Preview |
100.86[89.88, 112.74] | 64.49% | [2, 3] |
| 4 | Qwen3.7-Plus |
56.04[45.63, 66.45] | 58.12% | [4, 4] |
| 5 | Kimi-K3 |
9.62[-2.16, 21.71] | 51.32% | [5, 6] |
| 6 | Claude Opus 5 |
-2.47[-14.01, 8.96] | 49.51% | [5, 6] |
| 7 | Grok-4.5 |
-55.49[-68.68, -43.23] | 41.77% | [7, 8] |
| 8 | Claude Sonnet 5 |
-58.17[-68.74, -47.74] | 41.38% | [7, 8] |
| 9 | GLM-5V-Turbo |
-117.28[-129.38, -106.41] | 33.12% | [9, 9] |
| 10 | ERNIE-5.0 |
-190.68[-203.33, -178.59] | 23.88% | [10, 10] |