Vision & documents

OfficeQA Pro leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on OfficeQA Pro is 67.7% by Claude Opus 5.5. 96 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	GPT-6 Astra	67.7%	estimated ± 7.2 pp, medium confidence
2	GPT-6.1 Sol	67.7%	estimated ± 7.2 pp, medium confidence
3	Gemini 3.8 Flash	67.7%	estimated ± 7.2 pp, medium confidence
4	Gemini 3.7 Flash	67.7%	estimated ± 7.2 pp, medium confidence
5	Gemini 3.5 Flash	67.7%	estimated ± 7.2 pp, medium confidence
6	Claude Opus 5.5	67.7%	measured
7	GPT-5.6 Sol	67.7%	estimated ± 7.2 pp, medium confidence
8	Gemini 3.6 Flash	67.7%	estimated ± 7.2 pp, medium confidence
9	GPT-6 Sol	67.6%	estimated ± 7.2 pp, medium confidence
10	Qwen3.8 Max Preview	67.6%	estimated ± 7.2 pp, medium confidence
11	Gemini 3.1 Pro	67.4%	estimated ± 7.2 pp, medium confidence
12	Claude Opus 5	66.9%	measured
13	Claude Opus 4.8	66.2%	measured
14	Hy4 preview	66.2%	measured
15	Claude Sonnet 5.5	65.6%	measured
16	Kimi K3	63.3%	measured
17	GLM-5.3-Flash	62.4%	measured
18	GPT-5.6 Terra	62.0%	estimated ± 7.2 pp, medium confidence
19	Muse Spark	60.4%	estimated ± 7.2 pp, medium confidence
20	Qwen3.7 Plus	60.4%	estimated ± 7.2 pp, medium confidence
21	Step 5 Preview	60.3%	measured
22	Grok 4.5	59.6%	estimated ± 7.2 pp, medium confidence
23	Claude Fable 5	57.9%	measured
24	Gemini 3 Pro	57.8%	estimated ± 7.2 pp, medium confidence
25	Qwen3.8-Flash-Next	54.7%	estimated ± 7.2 pp, medium confidence
26	GPT-6 Luna	54.1%	estimated ± 7.2 pp, medium confidence
27	GPT-5.5	54.1%	measured
28	GPT-5.4	53.2%	measured
29	Kimi K2.6	52.6%	estimated ± 7.2 pp, medium confidence
30	Apodex 1.1	51.9%	estimated ± 7.2 pp, medium confidence
31	Apodex 1.1 Mini	51.9%	estimated ± 7.2 pp, medium confidence
32	Gemini 3.5 Flash-Lite	51.4%	estimated ± 7.2 pp, medium confidence
33	Ling 3.0 Flash VL	51.4%	estimated ± 7.2 pp, medium confidence
34	Gemini 3 Flash	50.8%	estimated ± 7.2 pp, medium confidence
35	GPT-5.6 Luna	50.8%	estimated ± 7.2 pp, medium confidence
36	GPT-5.3 Codex	50.7%	estimated ± 7.2 pp, medium confidence
37	Grok 4.3	50.5%	estimated ± 7.2 pp, medium confidence
38	Qwen3.6 Plus	50.5%	estimated ± 7.2 pp, medium confidence
39	Claude Sonnet 5	50.4%	estimated ± 7.2 pp, medium confidence
40	DeepSeek V4.1 Flash	50.3%	estimated ± 7.2 pp, medium confidence
41	Claude Opus 4.7	50.3%	estimated ± 7.2 pp, medium confidence
42	Mistral Large 4	50.3%	estimated ± 7.2 pp, medium confidence
43	GPT-5.2-Codex	50.3%	estimated ± 7.2 pp, low confidence
44	Qwen3.8-27B	50.3%	estimated ± 7.2 pp, low confidence
45	GPT-5.1	50.3%	estimated ± 7.2 pp, low confidence
46	Claude Opus 4.6 (Adaptive)	50.3%	estimated ± 7.2 pp, low confidence
47	Kimi K2.5	50.3%	estimated ± 7.2 pp, low confidence
48	Kimi K2.5 (Reasoning)	50.3%	estimated ± 7.2 pp, low confidence
49	Step 3.7 Flash	50.3%	estimated ± 7.2 pp, low confidence
50	Qwen3.5-122B-A10B	50.3%	estimated ± 7.2 pp, low confidence
51	Qwen3.5-27B	50.3%	estimated ± 7.2 pp, low confidence
52	Qwen3.6-35B-A3B	50.3%	estimated ± 7.2 pp, low confidence
53	Gemini 2.5 Pro	50.3%	estimated ± 7.2 pp, low confidence
54	Qwen3.6-27B	50.3%	estimated ± 7.2 pp, low confidence
55	Claude Opus 4.5 Thinking	50.3%	estimated ± 7.2 pp, low confidence
56	GPT-5 (high)	50.3%	estimated ± 7.2 pp, low confidence
57	GPT-5 (medium)	50.3%	estimated ± 7.2 pp, low confidence
58	Inkling-Small	50.3%	estimated ± 7.2 pp, low confidence
59	Muse Glimmer 30B	50.3%	estimated ± 7.2 pp, low confidence
60	Claude 3 Haiku	50.3%	estimated ± 7.2 pp, low confidence
61	Claude 4.1 Opus Thinking	50.3%	estimated ± 7.2 pp, low confidence
62	Claude 4 Sonnet	50.3%	estimated ± 7.2 pp, low confidence
63	Claude Opus 4.5	50.3%	estimated ± 7.2 pp, low confidence
64	Claude Opus 4.6	50.3%	estimated ± 7.2 pp, low confidence
65	Claude Sonnet 4.6	50.3%	estimated ± 7.2 pp, low confidence
66	Command A+	50.3%	estimated ± 7.2 pp, low confidence
67	Gemini 1.5 Pro	50.3%	estimated ± 7.2 pp, low confidence
68	Gemini 2.5 Flash	50.3%	estimated ± 7.2 pp, low confidence
69	Gemma 3 27B	50.3%	estimated ± 7.2 pp, low confidence
70	Gemma 4 12B	50.3%	estimated ± 7.2 pp, low confidence
71	Gemma 4 26B A4B	50.3%	estimated ± 7.2 pp, low confidence
72	Gemma 4 31B	50.3%	estimated ± 7.2 pp, low confidence
73	Gemma 4 E2B	50.3%	estimated ± 7.2 pp, low confidence
74	Gemma 4 E4B	50.3%	estimated ± 7.2 pp, low confidence
75	GLM-5V-Turbo	50.3%	estimated ± 7.2 pp, low confidence
76	GPT-4.1	50.3%	estimated ± 7.2 pp, low confidence
77	GPT-4.1 mini	50.3%	estimated ± 7.2 pp, low confidence
78	GPT-4.1 nano	50.3%	estimated ± 7.2 pp, low confidence
79	GPT-4o mini	50.3%	estimated ± 7.2 pp, low confidence
80	GPT-5.1-Codex	50.3%	estimated ± 7.2 pp, low confidence
81	GPT-5.1-Codex-Max	50.3%	estimated ± 7.2 pp, low confidence
82	GPT-5.4 mini	50.3%	estimated ± 7.2 pp, low confidence
83	GPT-5.4 nano	50.3%	estimated ± 7.2 pp, low confidence
84	Grok 4	50.3%	estimated ± 7.2 pp, low confidence
85	Grok 4.1 Fast	50.3%	estimated ± 7.2 pp, low confidence
86	Grok 4.1 Fast (Reasoning)	50.3%	estimated ± 7.2 pp, low confidence
87	Grok 4 Fast (Reasoning)	50.3%	estimated ± 7.2 pp, low confidence
88	Inkling	50.3%	estimated ± 7.2 pp, low confidence
89	LFM2.5-VL-1.6B-Extract	50.3%	estimated ± 7.2 pp, low confidence
90	Llama 4 Maverick	50.3%	estimated ± 7.2 pp, low confidence
91	Llama 4 Scout	50.3%	estimated ± 7.2 pp, low confidence
92	MiMo-V2.6-Flash	50.3%	estimated ± 7.2 pp, low confidence
93	MiMo-V2-Omni	50.3%	estimated ± 7.2 pp, low confidence
94	Mistral Large 3	50.3%	estimated ± 7.2 pp, low confidence
95	Mistral Medium 3	50.3%	estimated ± 7.2 pp, low confidence
96	Mistral Medium 3.5 128B	50.3%	estimated ± 7.2 pp, low confidence
97	Mistral Small 4	50.3%	estimated ± 7.2 pp, low confidence
98	Mistral Small 4 (Reasoning)	50.3%	estimated ± 7.2 pp, low confidence
99	Nemotron 3 Nano Omni 30B A3B	50.3%	estimated ± 7.2 pp, low confidence
100	Nova Pro	50.3%	estimated ± 7.2 pp, low confidence
101	o3	50.3%	estimated ± 7.2 pp, low confidence
102	Phi-4 Multimodal Instruct	50.3%	estimated ± 7.2 pp, low confidence
103	Qwen3.5-35B-A3B	50.3%	estimated ± 7.2 pp, low confidence
104	Qwen3.5 397B	50.3%	estimated ± 7.2 pp, low confidence
105	Qwen3.5 397B (Reasoning)	50.3%	estimated ± 7.2 pp, low confidence
106	Qwen3-Omni-30B-A3B-Instruct	50.3%	estimated ± 7.2 pp, low confidence
107	Qwen3-Omni-30B-A3B-Thinking	50.3%	estimated ± 7.2 pp, low confidence
108	MiniMax M3	45.1%	measured
109	Claude Opus 4.7 (Adaptive)	43.6%	measured

OfficeQA Pro leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General