Agentic · tools

Toolathlon-Verified leaderboard

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

As of 2026-10-07, the highest measured score on Toolathlon-Verified is 80.6% by Claude Opus 5. 143 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5	80.6%	measured
2	GPT-6 Astra	80.0%	estimated ± 1.0 pp, low confidence
3	Gemini 4 Argon	79.8%	estimated ± 1.0 pp, medium confidence
4	Muse Spark 1.3	79.7%	estimated ± 1.0 pp, medium confidence
5	GPT-5.6 Sol	79.5%	estimated ± 1.0 pp, medium confidence
6	GPT-6 Sol	79.4%	estimated ± 1.0 pp, medium confidence
7	Gemini 3.8 Flash	79.3%	estimated ± 1.0 pp, medium confidence
8	GPT-5.6 Terra	78.7%	estimated ± 1.0 pp, medium confidence
9	Gemini 3.7 Flash	78.5%	estimated ± 1.0 pp, medium confidence
10	GLM-5.3-Flash	78.4%	measured
11	GPT-5.6 Luna	78.3%	estimated ± 1.0 pp, medium confidence
12	Claude Fable 5.1	77.8%	measured
13	Claude Opus 5.5	77.8%	measured
14	Claude Sonnet 5.5	77.8%	measured
15	GPT-6.1 Sol	76.9%	estimated ± 1.9 pp, medium confidence
16	MiMo-V2.6-Pro	76.9%	measured
17	GPT-6 Luna	76.9%	estimated ± 2.2 pp, medium confidence
18	Claude Fable 5	76.7%	estimated ± 1.9 pp, medium confidence
19	Claude Haiku 5.5	76.6%	estimated ± 2.2 pp, medium confidence
20	Grok 4.7	76.4%	estimated ± 2.2 pp, medium confidence
21	Mistral Large 4	76.1%	estimated ± 2.2 pp, medium confidence
22	Grok 4.6	75.9%	estimated ± 2.3 pp, medium confidence
23	GPT-5.5 Pro	74.5%	estimated ± 7.7 pp, medium confidence
24	DeepSeek V4.1 Flash	74.4%	estimated ± 2.2 pp, medium confidence
25	Ling 3.1 Flash	74.3%	estimated ± 2.2 pp, low confidence
26	DeepSeek V4 Pro 0813	74.1%	measured
27	Hy4 preview	74.1%	measured
28	Step 5 Preview	74.1%	measured
29	Fugu Cyber	74.1%	estimated ± 2.2 pp, low confidence
30	Gemini 3.8 Flash Cyber	74.0%	estimated ± 2.2 pp, low confidence
31	Qwen3.8 Max Preview	74.0%	estimated ± 2.3 pp, medium confidence
32	GPT-5.4 Pro	73.7%	estimated ± 7.7 pp, medium confidence
33	MiMo-V2.6-Flash	73.6%	measured
34	Claude Mythos 5	73.5%	estimated ± 2.2 pp, low confidence
35	Qwen3.8-Flash-Next	73.5%	measured
36	Gemini 3.5 Flash Cyber	73.4%	estimated ± 2.2 pp, low confidence
37	Claude Opus 4.8	73.4%	estimated ± 1.0 pp, medium confidence
38	Claude Mythos Preview	73.4%	estimated ± 2.2 pp, low confidence
39	Kimi K3	73.2%	measured
40	Atria Dawn Preview	73.2%	estimated ± 1.3 pp, low confidence
41	Claude 4.1 Opus	73.2%	estimated ± 1.3 pp, low confidence
42	Claude 4 Sonnet	73.2%	estimated ± 1.3 pp, low confidence
43	Claude Haiku 4.5	73.2%	estimated ± 1.3 pp, low confidence
44	Claude Opus 4.5	73.2%	estimated ± 1.3 pp, low confidence
45	Claude Opus 4.6	73.2%	estimated ± 1.3 pp, low confidence
46	Claude Sonnet 4.5	73.2%	estimated ± 1.3 pp, low confidence
47	Gemini 3 Flash	73.2%	estimated ± 1.3 pp, low confidence
48	Gemini 3 Pro	73.2%	estimated ± 1.3 pp, low confidence
49	GPT-5.1-Codex	73.2%	estimated ± 1.3 pp, low confidence
50	GPT-5.2	73.2%	estimated ± 1.3 pp, low confidence
51	GPT-5.2-Codex	73.2%	estimated ± 1.3 pp, low confidence
52	GPT-5.3 Codex	73.2%	estimated ± 1.3 pp, low confidence
53	GPT-5.4	73.2%	estimated ± 1.3 pp, low confidence
54	GPT-5 (high)	73.2%	estimated ± 1.3 pp, low confidence
55	Kimi K2.5	73.2%	estimated ± 1.3 pp, low confidence
56	Qwen3.5 Plus	73.2%	estimated ± 1.3 pp, low confidence
57	Qwen3.8-27B	73.2%	estimated ± 1.3 pp, low confidence
58	GLM-5.3	73.0%	measured
59	Muse Glimmer 30B	73.0%	estimated ± 2.2 pp, low confidence
60	Qwen3.8 Max	72.5%	measured
61	Claude Opus 4.7 (Adaptive)	72.3%	estimated ± 1.0 pp, low confidence
62	Ornith-1.5-397B	71.2%	measured
63	Inkling	71.0%	estimated ± 1.9 pp, low confidence
64	Claude Sonnet 5	70.9%	estimated ± 2.3 pp, medium confidence
65	Muse Spark 1.2	70.7%	estimated ± 2.3 pp, medium confidence
66	DeepSeek V4 Flash 0731	70.3%	measured
67	GLM-5.1	70.2%	estimated ± 2.2 pp, low confidence
68	Muse Spark 1.1	69.8%	estimated ± 1.0 pp, low confidence
69	Claude Opus 4.7	69.6%	estimated ± 1.0 pp, low confidence
70	Grok 4.5	69.5%	estimated ± 2.3 pp, medium confidence
71	GPT-5.5	68.8%	estimated ± 1.0 pp, low confidence
72	GLM-5.2	67.6%	estimated ± 2.3 pp, medium confidence
73	Nemotron 3 Ultra	67.1%	estimated ± 1.9 pp, low confidence
74	Agents-A1	63.9%	estimated ± 5.9 pp, medium confidence
75	Claude Sonnet 4.6	62.8%	estimated ± 1.0 pp, low confidence
76	Quasar 438B	62.3%	estimated ± 2.3 pp, medium confidence
77	Beam	61.9%	estimated ± 7.7 pp, medium confidence
78	Muse Spark	61.4%	estimated ± 2.2 pp, low confidence
79	GLM-5	61.2%	estimated ± 2.2 pp, low confidence
80	Gemini 3.6 Flash	60.0%	estimated ± 2.3 pp, medium confidence
81	Apodex 1.1	59.6%	estimated ± 2.4 pp, high confidence
82	Apodex 1.1 Mini	59.6%	estimated ± 2.4 pp, high confidence
83	Ling 3.0 Flash VL	58.1%	estimated ± 2.4 pp, high confidence
84	Gemini 3.5 Flash	57.0%	estimated ± 2.3 pp, medium confidence
85	Qwen3.5 397B	57.0%	estimated ± 5.5 pp, low confidence
86	dots3-note Preview	55.6%	measured
87	Solar Pro 4	55.5%	estimated ± 2.4 pp, medium confidence
88	Hy3	55.3%	estimated ± 2.3 pp, medium confidence
89	Hy3 Preview	55.3%	estimated ± 2.3 pp, medium confidence
90	Inkling-Small	54.4%	measured
91	Qwen3.7 Max	53.2%	estimated ± 2.3 pp, low confidence
92	Kimi K2.6	52.5%	estimated ± 1.0 pp, low confidence
93	MiniMax M3	52.5%	estimated ± 1.0 pp, low confidence
94	MiMo-V2.5-Pro	51.7%	estimated ± 2.3 pp, low confidence
95	Kimi K2.7 Code	51.5%	estimated ± 2.3 pp, low confidence
96	Agents-A1-4B	51.4%	estimated ± 7.7 pp, medium confidence
97	GLM-4.7	50.4%	estimated ± 2.4 pp, medium confidence
98	Step 3.7 Flash	50.4%	estimated ± 2.4 pp, medium confidence
99	Laguna S 2.1	49.7%	measured
100	Ling 3.0 Flash	49.5%	estimated ± 2.3 pp, low confidence
101	Ling 3.0 Flash FP8	49.5%	estimated ± 2.3 pp, low confidence
102	Qwen3.6 Plus	49.1%	estimated ± 2.4 pp, medium confidence
103	Ornith-1.5-35B-A3B	48.7%	measured
104	Qwen3.6-27B	48.3%	estimated ± 2.3 pp, low confidence
105	GPT-5.4 mini	47.7%	estimated ± 2.3 pp, low confidence
106	A.X K2	47.3%	estimated ± 2.4 pp, medium confidence
107	Qwen3.5-27B	45.7%	estimated ± 7.7 pp, medium confidence
108	Qwen3.5-35B-A3B	45.7%	estimated ± 7.7 pp, medium confidence
109	GPT-5.4 nano	44.8%	estimated ± 2.3 pp, low confidence
110	Grok 4.3	44.0%	estimated ± 2.3 pp, low confidence
111	MiniMax M2.7	43.3%	estimated ± 2.3 pp, low confidence
112	Solar Open 2	42.8%	estimated ± 9.3 pp, medium confidence
113	Qwen3.7 Plus	42.5%	estimated ± 1.0 pp, low confidence
114	Gemini 3.5 Flash-Lite	41.9%	estimated ± 2.3 pp, low confidence
115	Ornith-1.5-9B	41.2%	measured
116	Qwen3.6-35B-A3B	40.3%	estimated ± 2.3 pp, low confidence
117	Kimi K2.5 (Reasoning)	38.9%	estimated ± 2.4 pp, medium confidence
118	GPT-5.1	37.8%	estimated ± 2.4 pp, medium confidence
119	LongCat-Flash-Lite-Sparse	33.4%	estimated ± 7.7 pp, low confidence
120	LLaDA2.2-flash	31.4%	estimated ± 9.3 pp, low confidence
121	Gemini 3.1 Pro	31.1%	estimated ± 2.3 pp, low confidence
122	Qwen3.5-122B-A10B	29.5%	estimated ± 2.3 pp, low confidence
123	Mistral Medium 3.5 128B	28.9%	estimated ± 2.3 pp, low confidence
124	MiniCPM5-2B	27.0%	estimated ± 2.4 pp, medium confidence
125	Gemma 4 31B	22.4%	estimated ± 2.3 pp, low confidence
126	GPT-OSS 120B	20.9%	estimated ± 2.3 pp, low confidence
127	Nemotron 3.5 Lightning 30B A3B NVFP4	20.8%	estimated ± 2.3 pp, low confidence
128	MiMo-V2-Flash	18.5%	estimated ± 2.4 pp, medium confidence
129	Nemotron 3 Super 100B	14.8%	estimated ± 2.3 pp, low confidence
130	Granite 4.2 8B	13.5%	estimated ± 2.3 pp, low confidence
131	Command A+	13.2%	estimated ± 2.3 pp, low confidence
132	Gemini 2.5 Pro	13.0%	estimated ± 2.3 pp, low confidence
133	Granite 4.2 30B	11.6%	estimated ± 2.4 pp, medium confidence
134	Gemma 4 26B A4B	10.3%	estimated ± 2.4 pp, medium confidence
135	Ling 3.0 Tiny	9.7%	estimated ± 2.4 pp, medium confidence
136	Mistral Large 3	9.2%	estimated ± 2.3 pp, low confidence
137	Mistral Small 4	5.6%	estimated ± 2.3 pp, low confidence
138	Mistral Small 4 (Reasoning)	5.6%	estimated ± 2.3 pp, low confidence
139	GPT-OSS 20B	5.5%	estimated ± 2.3 pp, low confidence
140	Trinity-Large-Preview	4.6%	estimated ± 2.3 pp, low confidence
141	Trinity-Large-Thinking	4.6%	estimated ± 2.3 pp, low confidence
142	Nemotron 3 Nano 30B	4.0%	estimated ± 2.3 pp, low confidence
143	DeepSeek V3	3.2%	estimated ± 2.3 pp, low confidence
144	Celeris-1	2.6%	estimated ± 2.3 pp, low confidence
145	Llama 4 Maverick	2.5%	estimated ± 2.3 pp, low confidence
146	Llama 4 Scout	2.2%	estimated ± 2.3 pp, low confidence
147	Gemma 3 27B	0.6%	estimated ± 2.3 pp, low confidence
148	DeepSeek V3 0324	0.0%	estimated ± 2.4 pp, medium confidence
149	Gemma 4 12B	0.0%	estimated ± 2.4 pp, medium confidence
150	Gemma 4 E2B	0.0%	estimated ± 2.4 pp, medium confidence
151	Gemma 4 E4B	0.0%	estimated ± 2.4 pp, medium confidence
152	GPT-4.1 mini	0.0%	estimated ± 2.4 pp, medium confidence
153	GPT-4.1 nano	0.0%	estimated ± 2.4 pp, medium confidence
154	GPT-4o	0.0%	estimated ± 2.4 pp, medium confidence
155	GPT-4o mini	0.0%	estimated ± 2.4 pp, medium confidence
156	Granite 4.2 3B	0.0%	estimated ± 2.4 pp, medium confidence
157	K-Exaone	0.0%	estimated ± 2.4 pp, medium confidence
158	LFM2.5-2.6B	0.0%	estimated ± 2.4 pp, medium confidence
159	Ling 2.6 Flash	0.0%	estimated ± 2.4 pp, medium confidence
160	Mercury 2.5	0.0%	estimated ± 2.4 pp, medium confidence
161	Nemotron 3 Nano Omni 30B A3B	0.0%	estimated ± 2.4 pp, medium confidence
162	North Mini Code	0.0%	estimated ± 2.4 pp, medium confidence
163	Solar Pro 3	0.0%	estimated ± 2.4 pp, medium confidence
164	Ultravox v0.6 Llama 3.3 70B	0.0%	estimated ± 2.4 pp, medium confidence

Toolathlon-Verified leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General