benchgap
Agentic · tools

Toolathlon leaderboard

As of 2026-11-21, the highest measured score on Toolathlon is 75.6% by Muse Spark 1.1. 178 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Ultramarine 1 Flash84.7%estimated ± 7.4 pp, low confidence
2Muse Spark 1.175.6%measured
3Qwen4 Max64.7%estimated ± 6.7 pp, low confidence
4Holo3-35B-A3B64.0%estimated ± 9.3 pp, medium confidence
5Atria Dawn Preview63.9%estimated ± 4.2 pp, medium confidence
6Kimi K3.1 Pro63.4%estimated ± 4.2 pp, high confidence
7LongCat-2.563.1%estimated ± 4.2 pp, high confidence
8Ling 3.1 Flash62.8%estimated ± 4.2 pp, high confidence
9GPT-6 Astra62.6%estimated ± 4.2 pp, high confidence
10Inkling 2 Small61.0%estimated ± 4.2 pp, high confidence
11GPT-5.5 Pro60.9%estimated ± 4.2 pp, high confidence
12Claude Opus 4.859.9%measured
13GPT-5.4 Pro59.9%estimated ± 4.2 pp, high confidence
14Claude Opus 559.3%estimated ± 1.7 pp, low confidence
15Step 5 Preview59.2%estimated ± 4.2 pp, high confidence
16Claude Fable 558.9%estimated ± 1.7 pp, low confidence
17Claude Mythos 558.3%estimated ± 4.2 pp, high confidence
18Holo3-122B-A10B58.1%estimated ± 9.3 pp, medium confidence
19GPT-5.6 Sol58.0%measured
20ERNIE 5.5 Flash Preview57.6%estimated ± 4.2 pp, high confidence
21Gemini 3.8 Flash56.7%estimated ± 1.7 pp, medium confidence
22Ornith-1.5-397B56.7%estimated ± 4.2 pp, high confidence
23Gemini 3.5 Flash56.5%measured
24GLM-5.3-Flash55.9%estimated ± 5.1 pp, low confidence
25Muse Wonder Full55.9%estimated ± 6.3 pp, low confidence
26Claude Sonnet 5.555.9%estimated ± 6.3 pp, low confidence
27Muse Wonder55.9%estimated ± 6.3 pp, low confidence
28MiMo-V2.6-Pro55.9%estimated ± 6.3 pp, low confidence
29Qwen3.8 Max Preview55.9%estimated ± 6.3 pp, low confidence
30Qwen3.8-Flash-Next55.9%estimated ± 6.3 pp, low confidence
31Gemini 4 Argon55.9%estimated ± 6.3 pp, low confidence
32Claude Haiku 5.555.9%estimated ± 6.3 pp, low confidence
33MiMo-V2.6-Flash55.9%estimated ± 6.3 pp, medium confidence
34GPT-6.1 Sol55.9%estimated ± 6.3 pp, medium confidence
35Muse Spark 1.255.9%estimated ± 6.3 pp, medium confidence
36MiniMax M3.1 Flash55.9%estimated ± 6.3 pp, medium confidence
37GPT-6 Luna55.9%estimated ± 6.3 pp, medium confidence
38Qwen3.8-27B55.9%estimated ± 6.3 pp, medium confidence
39Muse Spark Open55.9%estimated ± 6.3 pp, medium confidence
40Grok 4.555.9%estimated ± 6.3 pp, medium confidence
41Gemini 3.6 Flash55.8%estimated ± 6.3 pp, medium confidence
42Hy3 Preview55.8%estimated ± 6.3 pp, medium confidence
43Apodex 1.155.8%estimated ± 6.3 pp, medium confidence
44Apodex 1.1 Mini55.8%estimated ± 6.3 pp, medium confidence
45Quasar 438B55.8%estimated ± 6.3 pp, medium confidence
46Ling 3.0 Flash VL55.7%estimated ± 6.3 pp, medium confidence
47GPT-5.555.6%measured
48Muse Spark 1.355.6%estimated ± 1.7 pp, medium confidence
49Kimi K355.4%estimated ± 1.7 pp, medium confidence
50UI-Mate-27B55.3%estimated ± 9.3 pp, medium confidence
51Claude Fable 5.155.2%estimated ± 5.1 pp, low confidence
52GPT-6 Sol55.1%estimated ± 5.1 pp, low confidence
53Claude Sonnet 554.9%estimated ± 1.7 pp, medium confidence
54Gemini 3.7 Flash54.9%estimated ± 1.7 pp, medium confidence
55GPT-5.454.6%measured
56Mistral Large 454.3%estimated ± 4.2 pp, high confidence
57Grok 4.654.3%estimated ± 1.7 pp, medium confidence
58Grok 4.354.2%estimated ± 6.3 pp, medium confidence
59DeepSeek V4.1 Flash 101754.0%estimated ± 5.1 pp, low confidence
60GLM-5.353.6%estimated ± 5.1 pp, low confidence
61GPT-5.6 Luna53.4%measured
62MiniMax M353.2%estimated ± 4.2 pp, high confidence
63GPT-5.6 Terra53.1%measured
64dots3-note Preview53.0%estimated ± 4.2 pp, high confidence
65Claude Opus 4.753.0%estimated ± 1.7 pp, medium confidence
66Hy352.9%estimated ± 6.3 pp, medium confidence
67Hy4 preview52.5%estimated ± 5.2 pp, low confidence
68Claude Opus 4.652.5%estimated ± 1.7 pp, low confidence
69MiMo-V2.552.3%estimated ± 6.9 pp, medium confidence
70Claude Sonnet 4.651.9%estimated ± 1.7 pp, low confidence
71DeepSeek V4 Pro51.8%measured
72Qwen3.7 Max51.8%estimated ± 5.1 pp, low confidence
73Grok 4.751.6%estimated ± 5.1 pp, low confidence
74Sage-A1.551.3%estimated ± 4.2 pp, high confidence
75Kimi K2.7 Code50.2%estimated ± 6.3 pp, medium confidence
76Claude Opus 4.6 (Adaptive)50.2%estimated ± 6.7 pp, medium confidence
77Gemini 3 Pro50.1%estimated ± 7.2 pp, medium confidence
78Kimi K2.650.0%measured
79Ling 3.0 Flash Sante49.7%estimated ± 4.2 pp, high confidence
80MiMo-V2-Pro49.7%estimated ± 6.9 pp, medium confidence
81Step 3.7 Flash49.5%measured
82Qwen3.8 Max49.4%estimated ± 5.2 pp, low confidence
83Claude Opus 5.549.4%estimated ± 5.1 pp, low confidence
84MiMo-V2.5-Pro49.4%estimated ± 5.1 pp, low confidence
85Ripple-65B Preview49.1%estimated ± 7.5 pp, medium confidence
86Gemma 4 31B48.8%estimated ± 5.1 pp, low confidence
87Inkling 2 Mini48.5%estimated ± 4.2 pp, high confidence
88GLM-5.248.2%measured
89DeepSeek V4 Flash 073147.8%measured
90Nemotron 3.5 Super VL 120B A12B47.5%estimated ± 4.2 pp, high confidence
91Ling 3.0 Flash Fin47.1%estimated ± 6.7 pp, medium confidence
92Nemotron 3.5 Super VL 120B A12B NVFP447.0%estimated ± 4.2 pp, high confidence
93Inkling-Small46.9%estimated ± 4.2 pp, high confidence
94GPT-5.3 Codex46.6%estimated ± 7.2 pp, medium confidence
95Inkling46.6%estimated ± 4.2 pp, high confidence
96MiniMax M2.746.3%measured
97Beam 1.0 501B-A23B46.3%estimated ± 4.2 pp, high confidence
98Beam 1.0 501B-A23B FP846.2%estimated ± 4.2 pp, high confidence
99Gemini 3 Flash46.1%estimated ± 7.2 pp, medium confidence
100Beam 1.0 501B-A23B NVFP445.1%estimated ± 4.2 pp, high confidence
101Agents-A145.0%estimated ± 4.2 pp, high confidence
102Claude Opus 4.543.5%measured
103Muse Spark43.1%estimated ± 6.3 pp, medium confidence
104GPT-5.2-Codex43.0%estimated ± 7.2 pp, medium confidence
105GPT-5.4 mini42.9%measured
106Ling 3.0 Flash41.9%estimated ± 4.2 pp, high confidence
107GPT-5.1-Codex41.6%estimated ± 7.2 pp, medium confidence
108UI-Mate-9B41.3%estimated ± 9.3 pp, medium confidence
109Grok Build 0.141.2%estimated ± 7.2 pp, medium confidence
110Claude Sonnet 4.540.8%estimated ± 7.2 pp, medium confidence
111Qwen3.6-27B40.5%estimated ± 6.3 pp, medium confidence
112Solar Open 240.3%estimated ± 7.5 pp, medium confidence
113Gemini 3.5 Flash-Lite40.1%estimated ± 6.3 pp, medium confidence
114Grok 4.1 Fast40.0%estimated ± 7.2 pp, medium confidence
115Nemotron 3.5 Bolt 74B-A9B BF1640.0%estimated ± 4.2 pp, high confidence
116Qwen3.6 Plus39.8%measured
117GLM-5.138.2%estimated ± 4.2 pp, high confidence
118GLM-538.0%measured
119Ornith-1.5-35B-A3B37.9%estimated ± 4.2 pp, high confidence
120Qwen3 Max37.5%estimated ± 7.2 pp, medium confidence
121LLaDA2.2-flash37.2%estimated ± 7.5 pp, medium confidence
122Agents-A1-4B37.2%estimated ± 4.2 pp, high confidence
123Nemotron 3.5 Bolt 74B-A9B37.2%estimated ± 4.2 pp, high confidence
124Grok 436.6%estimated ± 7.2 pp, medium confidence
125GPT-5.236.4%estimated ± 4.2 pp, high confidence
126Qwen3.5 397B36.3%measured
127LFM3-1.7B36.3%estimated ± 7.5 pp, medium confidence
128GPT-5.4 nano35.5%measured
129A.X K235.3%estimated ± 6.3 pp, medium confidence
130Qwen3.5-122B-A10B34.8%estimated ± 4.2 pp, high confidence
131Claude 4 Sonnet34.7%estimated ± 7.2 pp, low confidence
132Gemini 3.1 Flash-Lite33.8%estimated ± 7.2 pp, low confidence
133Grok 4.2033.7%estimated ± 7.2 pp, low confidence
134Claude 4.1 Opus33.1%estimated ± 7.4 pp, low confidence
135Qwen3.5-27B32.6%estimated ± 4.2 pp, high confidence
136Qwen3.5-35B-A3B32.6%estimated ± 4.2 pp, high confidence
137Qwen3.5 Plus30.9%estimated ± 7.4 pp, low confidence
138GPT-530.0%estimated ± 6.3 pp, medium confidence
139Claude Haiku 4.529.7%estimated ± 7.4 pp, low confidence
140Ornith-1.5-9B29.2%estimated ± 4.2 pp, medium confidence
141GLM-5V-Turbo28.0%estimated ± 7.2 pp, low confidence
142Kimi K2.527.8%measured
143GPT-5.127.2%estimated ± 6.3 pp, low confidence
144K-EXAONE 2.027.1%estimated ± 6.3 pp, low confidence
145Gemini 3.1 Pro27.0%estimated ± 6.3 pp, low confidence
146DeepSeek V3.227.0%estimated ± 7.2 pp, low confidence
147Muse Glimmer 30B27.0%estimated ± 6.3 pp, low confidence
148Qwen3.7 Plus27.0%estimated ± 6.3 pp, low confidence
149Mistral Medium 3.5 128B27.0%estimated ± 6.3 pp, low confidence
150MiniCPM5-2B27.0%estimated ± 6.3 pp, low confidence
151MiMo-V2-Flash27.0%estimated ± 6.3 pp, low confidence
152GPT-OSS 120B27.0%estimated ± 6.3 pp, low confidence
153Granite 4.2 30B27.0%estimated ± 6.3 pp, low confidence
154Gemma 4 26B A4B27.0%estimated ± 6.3 pp, low confidence
155Ling 3.0 Tiny27.0%estimated ± 6.3 pp, low confidence
156Command A+27.0%estimated ± 6.3 pp, low confidence
157Celeris-127.0%estimated ± 6.3 pp, low confidence
158Claude 3.5 Haiku27.0%estimated ± 6.3 pp, low confidence
159DeepSeek V327.0%estimated ± 6.3 pp, low confidence
160DeepSeek V3 032427.0%estimated ± 6.3 pp, low confidence
161Gemini 2.5 Pro27.0%estimated ± 6.3 pp, low confidence
162Gemma 3 27B27.0%estimated ± 6.3 pp, low confidence
163Gemma 4 12B27.0%estimated ± 6.3 pp, low confidence
164Gemma 4 E2B27.0%estimated ± 6.3 pp, low confidence
165Gemma 4 E4B27.0%estimated ± 6.3 pp, low confidence
166GPT-4.1 mini27.0%estimated ± 6.3 pp, low confidence
167GPT-4.1 nano27.0%estimated ± 6.3 pp, low confidence
168GPT-4o27.0%estimated ± 6.3 pp, low confidence
169GPT-4o mini27.0%estimated ± 6.3 pp, low confidence
170GPT-OSS 20B27.0%estimated ± 6.3 pp, low confidence
171Granite 4.2 3B27.0%estimated ± 6.3 pp, low confidence
172Granite 4.2 8B27.0%estimated ± 6.3 pp, low confidence
173K-Exaone27.0%estimated ± 6.3 pp, low confidence
174LFM2.5-2.6B27.0%estimated ± 6.3 pp, low confidence
175Ling 2.6 Flash27.0%estimated ± 6.3 pp, low confidence
176Llama 3.3 70B27.0%estimated ± 6.3 pp, low confidence
177Llama 4 Maverick27.0%estimated ± 6.3 pp, low confidence
178Llama 4 Scout27.0%estimated ± 6.3 pp, low confidence
179Mercury 2.527.0%estimated ± 6.3 pp, low confidence
180Mistral Large 327.0%estimated ± 6.3 pp, low confidence
181Mistral Small 3.127.0%estimated ± 6.3 pp, low confidence
182Mistral Small 427.0%estimated ± 6.3 pp, low confidence
183Nemotron 3 Nano 30B27.0%estimated ± 6.3 pp, low confidence
184Nemotron 3 Nano Omni 30B A3B27.0%estimated ± 6.3 pp, low confidence
185North Mini Code27.0%estimated ± 6.3 pp, low confidence
186Solar Pro 327.0%estimated ± 6.3 pp, low confidence
187Trinity-Large-Preview27.0%estimated ± 6.3 pp, low confidence
188Trinity-Large-Thinking27.0%estimated ± 6.3 pp, low confidence
189Ultravox v0.6 Llama 3.3 70B27.0%estimated ± 6.3 pp, low confidence
190Qwen3.6-35B-A3B26.9%measured
191Spark-X2.6-4B26.8%estimated ± 4.2 pp, medium confidence
192GLM-4.726.1%estimated ± 4.2 pp, medium confidence
193Sarvam 105B24.4%estimated ± 4.2 pp, medium confidence
194Solar Pro 424.2%estimated ± 4.2 pp, medium confidence
195GPT-4.123.9%estimated ± 7.2 pp, low confidence
196LongCat-Flash-Lite-Sparse23.8%estimated ± 4.2 pp, medium confidence
197Nemotron 3 Ultra21.2%estimated ± 4.2 pp, medium confidence
198Grok 4.117.3%estimated ± 6.9 pp, low confidence
199Nemotron 3.5 Lightning 30B A3B NVFP416.7%estimated ± 4.2 pp, medium confidence
200Sarvam 30B16.0%estimated ± 4.2 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General