Agentic · tools

Claw-Eval leaderboard

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

As of 2026-10-07, the highest measured score on Claw-Eval is 81.4% by Ornith-1.5-397B. 116 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Fable 5.1	86.8%	estimated ± 13.6 pp, low confidence
2	GLM-5.3	85.3%	estimated ± 13.6 pp, low confidence
3	Grok 4.6	85.3%	estimated ± 13.6 pp, low confidence
4	Claude Fable 5	84.5%	estimated ± 13.6 pp, low confidence
5	Qwen3.8 Max Preview	84.0%	estimated ± 13.6 pp, low confidence
6	GPT-5.6 Sol	81.7%	estimated ± 7.9 pp, low confidence
7	GPT-6 Astra	81.7%	estimated ± 7.9 pp, low confidence
8	GPT-5.5 Pro	81.7%	estimated ± 7.9 pp, low confidence
9	GPT-5.4 Pro	81.7%	estimated ± 7.9 pp, low confidence
10	Claude Mythos 5	81.7%	estimated ± 7.9 pp, low confidence
11	GPT-5.6 Terra	81.7%	estimated ± 7.9 pp, low confidence
12	Muse Spark 1.2	81.6%	estimated ± 13.6 pp, low confidence
13	Ornith-1.5-397B	81.4%	measured
14	Grok 4.5	80.7%	estimated ± 13.6 pp, low confidence
15	Gemini 3.8 Flash	80.2%	estimated ± 13.6 pp, low confidence
16	Claude Sonnet 5	79.5%	estimated ± 7.9 pp, medium confidence
17	K-EXAONE 2.0	77.7%	measured
18	Gemini 3.7 Flash	77.6%	estimated ± 13.6 pp, low confidence
19	Ornith-1.0-397B	77.1%	measured
20	Quasar 438B	75.2%	estimated ± 13.6 pp, low confidence
21	Claude Opus 4.8	75.0%	estimated ± 3.2 pp, medium confidence
22	MiniMax M3	74.5%	measured
23	dots3-note Preview	73.4%	measured
24	Claude Opus 4.7	73.3%	estimated ± 3.2 pp, medium confidence
25	GLM-5.2	73.3%	estimated ± 3.2 pp, medium confidence
26	Gemini 3.6 Flash	73.3%	estimated ± 13.6 pp, low confidence
27	Claude Opus 5	73.1%	estimated ± 5.7 pp, low confidence
28	Ornith-1.5-35B-A3B	72.5%	measured
29	Qwen3.6-27B	72.4%	measured
30	Inkling-Small	71.1%	estimated ± 5.7 pp, medium confidence
31	Beam	70.8%	estimated ± 5.7 pp, medium confidence
32	Claude Opus 4.6	70.4%	measured
33	Kimi K2.7 Code	69.9%	estimated ± 5.7 pp, medium confidence
34	Ornith-1.0-35B	69.8%	measured
35	Muse Glimmer 30B	69.7%	estimated ± 5.7 pp, medium confidence
36	Hy3	69.3%	estimated ± 13.6 pp, low confidence
37	Inkling	69.3%	estimated ± 5.7 pp, medium confidence
38	DeepSeek V4 Pro 0813	69.1%	estimated ± 5.7 pp, medium confidence
39	Qwen3.6-35B-A3B	68.7%	measured
40	GPT-5.6 Luna	68.5%	estimated ± 7.9 pp, medium confidence
41	Claude Sonnet 4.6	67.8%	measured
42	DeepSeek V4 Flash 0731	67.6%	estimated ± 5.7 pp, medium confidence
43	Step 3.7 Flash	67.1%	measured
44	Ornith-1.5-9B	66.5%	measured
45	Claude Opus 4.7 (Adaptive)	66.2%	estimated ± 5.6 pp, low confidence
46	Hy4 preview	66.2%	estimated ± 5.6 pp, low confidence
47	Kimi K3	66.2%	estimated ± 5.6 pp, low confidence
48	MiMo-V2.6-Flash	66.2%	estimated ± 5.6 pp, low confidence
49	MiMo-V2.6-Pro	66.2%	estimated ± 5.6 pp, low confidence
50	Muse Spark 1.1	66.2%	estimated ± 5.6 pp, low confidence
51	Muse Spark 1.3	66.2%	estimated ± 5.6 pp, low confidence
52	Qwen3.8-Flash-Next	66.2%	estimated ± 5.6 pp, low confidence
53	Qwen3.8 Max	66.2%	estimated ± 5.6 pp, low confidence
54	Step 5 Preview	66.2%	estimated ± 5.6 pp, low confidence
55	GPT-5.2	66.0%	estimated ± 5.6 pp, low confidence
56	GPT-5.3 Codex	65.5%	estimated ± 5.6 pp, low confidence
57	Qwen3.7 Max	65.2%	measured
58	Solar Pro 4	65.1%	estimated ± 5.7 pp, medium confidence
59	Qwen3.8-27B	65.0%	estimated ± 5.6 pp, low confidence
60	BTL-3	64.9%	estimated ± 2.3 pp, low confidence
61	Atria Dawn Preview	64.7%	estimated ± 2.3 pp, low confidence
62	BTL-4	64.5%	estimated ± 2.3 pp, medium confidence
63	Ling 3.0 Flash	64.5%	estimated ± 2.3 pp, medium confidence
64	Pokee-Isaac 28B	64.4%	estimated ± 2.3 pp, medium confidence
65	LLaDA2.2-flash	64.2%	measured
66	Ling 3.0 Flash FP8	64.1%	estimated ± 13.6 pp, low confidence
67	Solar Open 2	64.1%	estimated ± 5.7 pp, medium confidence
68	MiniCPM5-2B	64.0%	estimated ± 2.3 pp, medium confidence
69	GPT-5.4 mini	63.9%	estimated ± 5.7 pp, medium confidence
70	MiMo-V2.5-Pro	63.8%	measured
71	Muse Spark	63.8%	measured
72	Gemini 3.5 Flash	63.7%	estimated ± 3.2 pp, high confidence
73	GPT-5.4 nano	63.4%	estimated ± 5.7 pp, medium confidence
74	Granite 4.2 30B	63.3%	estimated ± 2.3 pp, medium confidence
75	Ornith-1.0-9B	63.1%	measured
76	LFM2.5-2.6B	62.8%	measured
77	Agents-A1-4B	62.8%	estimated ± 7.1 pp, low confidence
78	Agents-A1	62.7%	estimated ± 7.1 pp, low confidence
79	Qwen3.7 Plus	62.7%	measured
80	Nemotron 3.5 Lightning 30B A3B NVFP4	62.4%	estimated ± 7.9 pp, low confidence
81	Nemotron 3 Ultra	62.4%	estimated ± 7.9 pp, low confidence
82	Qwen3.5-122B-A10B	62.4%	estimated ± 7.9 pp, medium confidence
83	GLM-5.1	62.3%	measured
84	Kimi K2.6	62.3%	measured
85	MiMo-V2.5	62.3%	measured
86	GPT-5.5	61.0%	estimated ± 3.2 pp, high confidence
87	Granite 4.2 3B	60.6%	estimated ± 2.3 pp, medium confidence
88	Granite 4.2 8B	60.6%	estimated ± 2.3 pp, medium confidence
89	GPT-5.4	60.3%	measured
90	LongCat-Flash-Lite-Sparse	60.0%	estimated ± 5.7 pp, medium confidence
91	Claude Opus 4.5	59.6%	measured
92	Grok Build 0.1	59.4%	estimated ± 6.8 pp, low confidence
93	LFM2.5-8B-A1B	59.1%	estimated ± 2.3 pp, medium confidence
94	Qwen3.6 Plus	58.8%	measured
95	Grok 4.1 Fast	58.7%	estimated ± 6.8 pp, low confidence
96	Gemini 3.1 Pro	57.8%	measured
97	MiMo-V2-Pro	57.8%	measured
98	GLM-5	57.7%	measured
99	LLaDA2.2-mini	57.2%	measured
100	Qwen3 Max	57.1%	estimated ± 6.8 pp, low confidence
101	Qwen3.5 397B	56.8%	measured
102	Gemini 3.5 Flash-Lite	56.8%	estimated ± 13.6 pp, low confidence
103	Grok 4	56.5%	estimated ± 6.8 pp, low confidence
104	Gemini 2.5 Pro	56.3%	estimated ± 6.8 pp, low confidence
105	GPT-5.1	55.9%	estimated ± 6.8 pp, low confidence
106	GLM-5-Turbo	55.8%	measured
107	Mellum2-12B-A2.5B-Thinking	55.6%	estimated ± 2.3 pp, low confidence
108	Grok 4.1	55.4%	estimated ± 3.2 pp, high confidence
109	GLM-4.7	55.3%	estimated ± 6.8 pp, low confidence
110	Qwen3.5-27B	55.0%	estimated ± 6.8 pp, low confidence
111	Mistral Medium 3.5 128B	54.8%	estimated ± 6.8 pp, low confidence
112	Grok 4.3	54.5%	estimated ± 3.2 pp, medium confidence
113	Gemini 3.1 Flash-Lite	54.5%	estimated ± 6.8 pp, low confidence
114	Grok 4.20	54.5%	estimated ± 6.8 pp, low confidence
115	Mellum2-12B-A2.5B-Instruct	53.9%	estimated ± 2.3 pp, low confidence
116	GLM-5V-Turbo	53.8%	measured
117	Hy3 Preview	53.6%	estimated ± 6.8 pp, low confidence
118	Gemma 4 31B	52.7%	estimated ± 6.8 pp, low confidence
119	Kimi K2.5	52.3%	measured
120	Kimi K2.5 (Reasoning)	51.0%	estimated ± 6.8 pp, low confidence
121	Trinity-Large-Thinking	51.0%	estimated ± 6.8 pp, low confidence
122	Claude Sonnet 4.5	50.8%	estimated ± 5.6 pp, low confidence
123	GPT-5.1-Codex	50.7%	estimated ± 5.6 pp, low confidence
124	GPT-5.2-Codex	50.7%	estimated ± 5.6 pp, low confidence
125	Claude 4.1 Opus	50.7%	estimated ± 5.6 pp, low confidence
126	Claude 4 Sonnet	50.7%	estimated ± 5.6 pp, low confidence
127	Claude Haiku 4.5	50.7%	estimated ± 5.6 pp, low confidence
128	Gemini 3 Pro	50.7%	estimated ± 5.6 pp, low confidence
129	GPT-5 (high)	50.7%	estimated ± 5.6 pp, low confidence
130	Qwen3.5 Plus	50.7%	estimated ± 5.6 pp, low confidence
131	Gemini 3 Flash	49.2%	measured
132	GPT-OSS 120B	48.9%	estimated ± 6.8 pp, low confidence
133	MiniMax M2.7	48.7%	measured
134	Qwen3.5-35B-A3B	48.5%	estimated ± 6.8 pp, low confidence
135	GPT-4.1	45.8%	estimated ± 6.8 pp, low confidence
136	ZAYA1-8B	45.7%	estimated ± 2.3 pp, low confidence
137	MiMo-V2-Omni	45.2%	measured
138	DeepSeek V3.2	40.2%	measured
139	LFM2.5-VL-3B	28.1%	estimated ± 2.3 pp, low confidence
140	Command A+	21.7%	estimated ± 13.6 pp, low confidence
141	Mistral Large 3	15.6%	estimated ± 13.6 pp, low confidence
142	Mistral Small 4	9.8%	estimated ± 13.6 pp, low confidence
143	Mistral Small 4 (Reasoning)	9.8%	estimated ± 13.6 pp, low confidence
144	GPT-OSS 20B	9.5%	estimated ± 13.6 pp, low confidence
145	MiniCPM5-1B	9.1%	estimated ± 2.3 pp, low confidence
146	Trinity-Large-Preview	8.0%	estimated ± 13.6 pp, low confidence
147	Nemotron 3 Nano 30B	7.0%	estimated ± 13.6 pp, low confidence
148	DeepSeek V3	5.6%	estimated ± 13.6 pp, low confidence
149	Nemotron 3 Super 100B	5.5%	measured
150	Celeris-1	4.6%	estimated ± 13.6 pp, low confidence
151	Llama 4 Maverick	4.4%	estimated ± 13.6 pp, low confidence
152	Llama 4 Scout	4.0%	estimated ± 13.6 pp, low confidence
153	LFM2.5-VL-450M	3.5%	estimated ± 2.3 pp, low confidence
154	LFM2.5-230M	3.4%	estimated ± 2.3 pp, low confidence
155	Gemma 3 27B	1.0%	estimated ± 13.6 pp, low confidence

Claw-Eval leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General