Agentic · tools

AA Harvey LAB leaderboard

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

As of 2026-10-07, the highest measured score on AA Harvey LAB is 94.6% by Kimi K3. 99 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Kimi K3	94.6%	measured
2	Muse Spark 1.3	93.7%	estimated ± 1.2 pp, medium confidence
3	GLM-5.3	93.6%	estimated ± 1.2 pp, medium confidence
4	Grok 4.6	93.6%	estimated ± 1.2 pp, medium confidence
5	Claude Fable 5	93.6%	measured
6	GPT-6 Astra	93.5%	estimated ± 1.2 pp, medium confidence
7	Atria Dawn Preview	93.5%	estimated ± 9.8 pp, low confidence
8	Claude Opus 5	93.5%	measured
9	GPT-5.6 Sol	93.5%	estimated ± 1.2 pp, medium confidence
10	Qwen3.8 Max Preview	93.4%	estimated ± 1.2 pp, medium confidence
11	DeepSeek V4 Pro 0813	93.4%	estimated ± 1.2 pp, medium confidence
12	Step 5 Preview	93.4%	measured
13	Gemini 4 Argon	93.3%	estimated ± 1.3 pp, medium confidence
14	Qwen3.8-27B	93.2%	estimated ± 1.2 pp, medium confidence
15	Claude Sonnet 5.5	93.1%	measured
16	Claude Fable 5.1	93.0%	measured
17	GPT-5.5 Pro	92.9%	estimated ± 9.8 pp, low confidence
18	GPT-6.1 Sol	92.9%	estimated ± 2.1 pp, high confidence
19	Claude Sonnet 5	92.9%	estimated ± 1.2 pp, medium confidence
20	Muse Spark 1.2	92.9%	estimated ± 1.2 pp, medium confidence
21	GPT-5.6 Terra	92.9%	estimated ± 1.2 pp, medium confidence
22	GPT-5.4 Pro	92.8%	estimated ± 9.8 pp, low confidence
23	GPT-5.6 Luna	92.7%	estimated ± 1.2 pp, medium confidence
24	Claude Opus 4.8	92.7%	estimated ± 1.2 pp, medium confidence
25	GPT-6 Sol	92.7%	estimated ± 1.3 pp, medium confidence
26	Grok 4.5	92.6%	estimated ± 1.2 pp, medium confidence
27	DeepSeek V4 Flash 0731	92.5%	estimated ± 1.2 pp, medium confidence
28	Claude Mythos 5	92.4%	estimated ± 9.8 pp, low confidence
29	Gemini 3.8 Flash	92.4%	estimated ± 1.2 pp, medium confidence
30	Claude Opus 4.7 (Adaptive)	92.1%	estimated ± 1.2 pp, medium confidence
31	GLM-5.2	92.0%	estimated ± 1.2 pp, medium confidence
32	DeepSeek V4.1 Flash	92.0%	estimated ± 1.7 pp, medium confidence
33	GPT-5.5	91.4%	estimated ± 1.2 pp, medium confidence
34	Claude Opus 5.5	91.2%	measured
35	Gemini 3.7 Flash	90.7%	measured
36	Claude Haiku 5.5	89.9%	measured
37	Beam	89.9%	estimated ± 9.8 pp, low confidence
38	Apodex 1.1	89.8%	estimated ± 1.7 pp, medium confidence
39	Ornith-1.5-397B	89.8%	estimated ± 1.7 pp, medium confidence
40	Qwen3.8-Flash-Next	89.7%	estimated ± 1.3 pp, medium confidence
41	Qwen3.8 Max	89.7%	estimated ± 1.3 pp, medium confidence
42	Hy4 preview	89.6%	estimated ± 1.7 pp, medium confidence
43	Quasar 438B	89.4%	estimated ± 1.2 pp, medium confidence
44	Claude Opus 4.7	89.3%	estimated ± 1.3 pp, medium confidence
45	Claude Sonnet 4.6	88.9%	estimated ± 1.3 pp, medium confidence
46	GLM-5.3-Flash	88.8%	estimated ± 1.3 pp, medium confidence
47	dots3-note Preview	88.7%	estimated ± 1.7 pp, medium confidence
48	MiniMax M3	88.4%	measured
49	Gemini 3.6 Flash	87.7%	estimated ± 1.2 pp, medium confidence
50	Agents-A1-4B	87.2%	estimated ± 9.8 pp, low confidence
51	GPT-5.2	86.9%	estimated ± 9.8 pp, low confidence
52	Agents-A1	86.8%	estimated ± 1.7 pp, medium confidence
53	Step 3.7 Flash	86.6%	estimated ± 1.7 pp, medium confidence
54	Qwen3.5 397B	86.0%	estimated ± 9.8 pp, low confidence
55	Muse Spark 1.1	85.8%	estimated ± 1.2 pp, medium confidence
56	Qwen3.5-27B	85.7%	estimated ± 9.8 pp, low confidence
57	Qwen3.5-35B-A3B	85.7%	estimated ± 9.8 pp, low confidence
58	Kimi K2.5	85.6%	estimated ± 9.8 pp, low confidence
59	Kimi K2.5 (Reasoning)	85.6%	estimated ± 9.8 pp, low confidence
60	Gemini 3.5 Flash	85.6%	estimated ± 1.2 pp, medium confidence
61	Hy3	84.3%	estimated ± 1.2 pp, medium confidence
62	Hy3 Preview	84.3%	estimated ± 1.2 pp, medium confidence
63	GLM-5.1	84.0%	estimated ± 1.2 pp, medium confidence
64	Inkling-Small	83.8%	estimated ± 1.2 pp, medium confidence
65	Inkling	83.4%	estimated ± 1.2 pp, medium confidence
66	GLM-4.7	83.3%	estimated ± 9.8 pp, low confidence
67	Qwen3.7 Max	83.1%	estimated ± 1.2 pp, medium confidence
68	Solar Pro 4	82.6%	estimated ± 9.8 pp, low confidence
69	LongCat-Flash-Lite-Sparse	82.4%	estimated ± 9.8 pp, low confidence
70	MiMo-V2.5-Pro	82.3%	estimated ± 1.2 pp, medium confidence
71	Kimi K2.7 Code	82.2%	estimated ± 1.2 pp, medium confidence
72	Kimi K2.6	82.0%	estimated ± 1.2 pp, medium confidence
73	Nemotron 3 Ultra	81.7%	measured
74	Ling 3.0 Flash	81.4%	estimated ± 1.2 pp, low confidence
75	Ling 3.0 Flash FP8	81.4%	estimated ± 1.2 pp, low confidence
76	Qwen3.6-27B	81.0%	estimated ± 1.2 pp, low confidence
77	Qwen3.7 Plus	80.8%	estimated ± 1.2 pp, low confidence
78	GPT-5.4 mini	80.8%	estimated ± 1.2 pp, low confidence
79	GPT-5.4 nano	80.2%	estimated ± 1.2 pp, low confidence
80	Grok 4.3	80.1%	estimated ± 1.2 pp, low confidence
81	MiniMax M2.7	80.0%	estimated ± 1.2 pp, low confidence
82	Gemini 3.5 Flash-Lite	79.8%	estimated ± 1.2 pp, low confidence
83	Qwen3.6-35B-A3B	79.7%	estimated ± 1.2 pp, low confidence
84	Muse Glimmer 30B	79.4%	estimated ± 1.2 pp, low confidence
85	Gemini 3.1 Pro	79.4%	estimated ± 1.2 pp, low confidence
86	Qwen3.5-122B-A10B	79.4%	estimated ± 1.2 pp, low confidence
87	Mistral Medium 3.5 128B	79.4%	estimated ± 1.2 pp, low confidence
88	Gemma 4 31B	79.4%	estimated ± 1.2 pp, low confidence
89	GPT-OSS 120B	79.4%	estimated ± 1.2 pp, low confidence
90	Nemotron 3.5 Lightning 30B A3B NVFP4	79.4%	estimated ± 1.2 pp, low confidence
91	Nemotron 3 Super 100B	79.4%	estimated ± 1.2 pp, low confidence
92	Command A+	79.4%	estimated ± 1.2 pp, low confidence
93	Gemini 2.5 Pro	79.4%	estimated ± 1.2 pp, low confidence
94	Granite 4.2 8B	79.4%	estimated ± 1.2 pp, low confidence
95	Celeris-1	79.4%	estimated ± 1.2 pp, low confidence
96	DeepSeek V3	79.4%	estimated ± 1.2 pp, low confidence
97	Gemma 3 27B	79.4%	estimated ± 1.2 pp, low confidence
98	GPT-OSS 20B	79.4%	estimated ± 1.2 pp, low confidence
99	Llama 4 Maverick	79.4%	estimated ± 1.2 pp, low confidence
100	Llama 4 Scout	79.4%	estimated ± 1.2 pp, low confidence
101	Mistral Large 3	79.4%	estimated ± 1.2 pp, low confidence
102	Mistral Small 4	79.4%	estimated ± 1.2 pp, low confidence
103	Mistral Small 4 (Reasoning)	79.4%	estimated ± 1.2 pp, low confidence
104	Nemotron 3 Nano 30B	79.4%	estimated ± 1.2 pp, low confidence
105	Trinity-Large-Preview	79.4%	estimated ± 1.2 pp, low confidence
106	Trinity-Large-Thinking	79.4%	estimated ± 1.2 pp, low confidence
107	Ornith-1.5-35B-A3B	79.2%	estimated ± 1.7 pp, low confidence
108	Ornith-1.5-9B	77.1%	estimated ± 1.7 pp, low confidence
109	Grok 4.7	19.6%	measured
110	GPT-5.4	0.1%	estimated ± 2.0 pp, low confidence
111	Claude Opus 4.6	0.0%	estimated ± 2.0 pp, low confidence

AA Harvey LAB leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General