Agentic · terminal

Terminal-Bench 3.0 leaderboard

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

As of 2026-10-07, the highest measured score on Terminal-Bench 3.0 is 42.7% by Claude Opus 5. 83 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5	42.7%	measured
2	GPT-6 Astra	38.5%	estimated ± 6.9 pp, low confidence
3	SWE-2	36.4%	estimated ± 9.1 pp, low confidence
4	GPT-5.6 Sol	34.6%	measured
5	Claude Fable 5.1	34.5%	estimated ± 6.9 pp, medium confidence
6	Claude Fable 5	34.1%	measured
7	Gemini 3.8 Flash	28.3%	estimated ± 6.9 pp, medium confidence
8	Kimi K3	27.6%	estimated ± 6.9 pp, medium confidence
9	MiMo-V2.6-Pro	27.0%	estimated ± 9.1 pp, low confidence
10	Grok 4.6	26.5%	measured
11	Claude Mythos 5	23.0%	estimated ± 9.1 pp, low confidence
12	MiMo-V2.6-Flash	22.2%	estimated ± 9.1 pp, low confidence
13	Claude Opus 4.8	21.1%	measured
14	GPT-5.5	21.0%	estimated ± 6.9 pp, medium confidence
15	GPT-5.6 Terra	20.8%	measured
16	DeepSeek V4.1 Flash	18.5%	estimated ± 6.9 pp, medium confidence
17	Step 5 Preview	18.3%	estimated ± 9.1 pp, low confidence
18	Gemini 3.5 Flash	18.2%	estimated ± 6.9 pp, medium confidence
19	Gemini 3.6 Flash	17.7%	estimated ± 6.9 pp, medium confidence
20	Grok 4.7	17.2%	estimated ± 6.9 pp, medium confidence
21	Muse Spark 1.3	15.9%	estimated ± 6.9 pp, medium confidence
22	Grok 4.5	15.7%	measured
23	Sakana Fugu-Ultra	15.2%	estimated ± 9.1 pp, low confidence
24	Ember-1	15.1%	estimated ± 9.1 pp, low confidence
25	GLM-5.3	15.0%	estimated ± 6.9 pp, medium confidence
26	Gemini 3.7 Flash	14.9%	measured
27	SWE-1.7	14.6%	estimated ± 9.1 pp, low confidence
28	Claude Sonnet 5	14.6%	measured
29	GPT-5.6 Luna	14.3%	measured
30	Gemini 3.1 Pro	14.2%	estimated ± 6.9 pp, medium confidence
31	Sakana Fugu	13.6%	estimated ± 9.1 pp, low confidence
32	Ornith-1.5-397B	13.5%	measured
33	Beam	13.5%	estimated ± 9.1 pp, low confidence
34	Muse Spark 1.2	13.1%	estimated ± 6.9 pp, medium confidence
35	Muse Spark 1.1	12.7%	estimated ± 6.9 pp, medium confidence
36	Atria Dawn Preview	12.2%	estimated ± 9.1 pp, low confidence
37	Claude Opus 4.7	11.9%	estimated ± 6.9 pp, medium confidence
38	Ornith-1.0-397B	11.7%	estimated ± 9.1 pp, low confidence
39	Qwen3.8 Max	10.9%	estimated ± 6.9 pp, low confidence
40	DeepSeek V4 Flash 0731	10.6%	estimated ± 6.9 pp, low confidence
41	Kimi K2.7 Code	10.6%	estimated ± 6.9 pp, low confidence
42	dots3-note Preview	10.3%	estimated ± 9.1 pp, low confidence
43	Seed 2.1 Pro	8.5%	estimated ± 9.1 pp, low confidence
44	Apodex 1.1	8.4%	estimated ± 9.1 pp, low confidence
45	Laguna S 2.1	8.2%	estimated ± 9.1 pp, low confidence
46	Quasar 438B	7.8%	estimated ± 9.1 pp, low confidence
47	GLM-5.3-Flash	7.4%	estimated ± 6.9 pp, low confidence
48	Seed 2.1 Turbo	7.3%	estimated ± 9.1 pp, low confidence
49	Pokee-Isaac 28B	6.5%	estimated ± 9.1 pp, low confidence
50	Ornith-1.0-35B	6.3%	estimated ± 9.1 pp, low confidence
51	Qwen3.7 Max	6.3%	estimated ± 6.9 pp, low confidence
52	MiMo-V2.5	6.1%	estimated ± 6.9 pp, low confidence
53	MAI-Code-1.1-Flash	6.0%	estimated ± 9.1 pp, low confidence
54	Step 3.7 Flash	5.2%	estimated ± 9.1 pp, low confidence
55	Ornith-1.5-35B-A3B	5.1%	measured
56	Composer 2.5	4.9%	estimated ± 6.9 pp, low confidence
57	Qwen3.8-27B	4.9%	estimated ± 6.9 pp, low confidence
58	Solar Pro 4	4.7%	estimated ± 9.1 pp, low confidence
59	GLM-5.2	4.6%	measured
60	Claude Sonnet 4.6	4.4%	estimated ± 6.9 pp, low confidence
61	MiMo-V2.5-Pro	4.4%	estimated ± 6.9 pp, low confidence
62	GLM-5.1	4.2%	estimated ± 6.9 pp, low confidence
63	Ternary Bonsai 2 27B	4.0%	estimated ± 9.1 pp, low confidence
64	Muse Glimmer 30B	3.8%	estimated ± 9.1 pp, low confidence
65	Hy4 preview	3.5%	estimated ± 6.9 pp, low confidence
66	Inkling-Small	3.5%	estimated ± 6.9 pp, low confidence
67	DeepSeek V4 Pro 0813	3.3%	estimated ± 6.9 pp, low confidence
68	GPT-5.4 mini	3.3%	estimated ± 6.9 pp, low confidence
69	Ornith-1.5-9B	3.1%	estimated ± 9.1 pp, low confidence
70	Gemini 3 Flash	3.1%	estimated ± 6.9 pp, low confidence
71	Kimi K2.6	3.0%	estimated ± 6.9 pp, low confidence
72	MiniMax M3	3.0%	estimated ± 6.9 pp, low confidence
73	Qwen3.6 Plus	2.8%	estimated ± 6.9 pp, low confidence
74	K-EXAONE 2.0	2.8%	estimated ± 9.1 pp, low confidence
75	Ornith-1.0-9B	2.7%	estimated ± 9.1 pp, low confidence
76	Qwen3.7 Plus	2.7%	estimated ± 6.9 pp, low confidence
77	Nemotron 3 Ultra	2.2%	estimated ± 6.9 pp, low confidence
78	A.X K2	2.0%	estimated ± 9.1 pp, low confidence
79	Gemini 3.5 Flash-Lite	2.0%	estimated ± 6.9 pp, low confidence
80	Ling 3.0 Flash	2.0%	estimated ± 6.9 pp, low confidence
81	MiniMax M2.7	1.7%	estimated ± 6.9 pp, low confidence
82	Granite 4.2 30B	1.5%	estimated ± 9.1 pp, low confidence
83	Inkling	1.5%	estimated ± 6.9 pp, low confidence
84	Nemotron 3.5 Lightning 30B A3B NVFP4	1.1%	estimated ± 9.1 pp, low confidence
85	Grok 4.20	0.9%	estimated ± 6.9 pp, low confidence
86	Granite 4.2 8B	0.9%	estimated ± 9.1 pp, low confidence
87	Claude Haiku 4.5	0.9%	estimated ± 6.9 pp, low confidence
88	Grok 4.3	0.7%	estimated ± 6.9 pp, low confidence
89	GPT-5.4 nano	0.7%	estimated ± 6.9 pp, low confidence
90	Mistral Medium 3.5 128B	0.4%	estimated ± 6.9 pp, low confidence
91	MiniCPM5-2B	0.3%	estimated ± 9.1 pp, low confidence
92	Mercury 2.5	0.2%	estimated ± 6.9 pp, low confidence
93	Gemini 3.1 Flash-Lite	0.2%	estimated ± 6.9 pp, low confidence
94	Laguna M.1	0.2%	estimated ± 6.9 pp, low confidence
95	Laguna XS.2	0.0%	estimated ± 6.9 pp, low confidence
96	Command A+	0.0%	estimated ± 6.9 pp, low confidence

Terminal-Bench 3.0 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General