Agentic · terminal

AA Terminal-Bench 4.0 leaderboard

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

As of 2026-10-07, the highest measured score on AA Terminal-Bench 4.0 is 63.6% by Claude Sonnet 5.5. 84 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Sonnet 5.5	63.6%	measured
2	Claude Opus 5.5	59.6%	measured
3	GPT-6 Astra	59.1%	measured
4	Gemini 4 Argon	57.1%	measured
5	Claude Mythos 5.1	56.5%	estimated ± 4.7 pp, high confidence
6	GPT-6.1 Sol	56.1%	measured
7	Claude Fable 5.1	52.0%	measured
8	Pareto 26.9	47.1%	estimated ± 4.7 pp, high confidence
9	Pareto 26.10 Preview	47.0%	estimated ± 4.7 pp, high confidence
10	GPT-6 Sol	43.9%	measured
11	GLM-5.3	41.9%	measured
12	Claude Opus 5	41.5%	estimated ± 14.0 pp, low confidence
13	Ling 3.1 Flash	37.1%	estimated ± 4.7 pp, high confidence
14	MiMo-V2.6-Pro	34.8%	measured
15	Grok 4.6	34.4%	estimated ± 14.0 pp, low confidence
16	GPT-5.6 Sol	33.8%	estimated ± 8.9 pp, low confidence
17	Muse Spark 1.3	33.3%	measured
18	Step 5 Preview	33.3%	measured
19	Claude Haiku 5.5	32.8%	measured
20	GLM-5.3-Flash	32.8%	measured
21	GPT-5.5	32.2%	estimated ± 14.0 pp, low confidence
22	Gemini 3.6 Flash	29.3%	estimated ± 14.0 pp, low confidence
23	Claude Mythos 5	28.6%	estimated ± 8.9 pp, medium confidence
24	DeepSeek V4 Pro 0813	28.5%	estimated ± 8.9 pp, medium confidence
25	GPT-5.6 Terra	27.8%	estimated ± 8.9 pp, medium confidence
26	DeepSeek V4.1 Flash	26.8%	measured
27	Mistral Large 4	26.8%	measured
28	Qwen3.8 Max	26.8%	estimated ± 8.9 pp, medium confidence
29	MiMo-V2.6-Flash	26.2%	estimated ± 4.7 pp, high confidence
30	Ornith-1.5-397B	26.1%	estimated ± 8.9 pp, medium confidence
31	Gemini 3.1 Pro	25.9%	estimated ± 14.0 pp, low confidence
32	Grok 4.7	25.8%	measured
33	Gemini 3.7 Flash	25.7%	estimated ± 8.9 pp, medium confidence
34	Hy4 preview	25.2%	estimated ± 8.9 pp, medium confidence
35	SWE-2	24.8%	estimated ± 4.7 pp, high confidence
36	GPT-5.6 Luna	24.3%	estimated ± 8.9 pp, medium confidence
37	Claude Fable 5	23.8%	estimated ± 8.9 pp, medium confidence
38	Claude Opus 4.7	23.3%	estimated ± 14.0 pp, low confidence
39	Grok 4.5	22.6%	estimated ± 8.9 pp, medium confidence
40	Muse Spark 1.2	22.1%	estimated ± 8.9 pp, medium confidence
41	DeepSeek V4 Flash 0731	21.8%	estimated ± 8.9 pp, medium confidence
42	Kimi K2.7 Code	21.6%	estimated ± 14.0 pp, low confidence
43	Sakana Fugu-Ultra	21.1%	estimated ± 8.9 pp, medium confidence
44	Ember-1	21.0%	estimated ± 8.9 pp, medium confidence
45	SWE-1.7	20.4%	estimated ± 8.9 pp, medium confidence
46	GLM-5.2	19.8%	estimated ± 8.9 pp, medium confidence
47	Gemini 3.8 Flash	19.7%	measured
48	Claude Sonnet 5	19.0%	estimated ± 8.9 pp, medium confidence
49	Sakana Fugu	18.8%	estimated ± 8.9 pp, medium confidence
50	Beam	18.7%	estimated ± 8.9 pp, medium confidence
51	Muse Spark 1.1	18.5%	estimated ± 8.9 pp, medium confidence
52	Atria Dawn Preview	16.5%	estimated ± 8.9 pp, medium confidence
53	Ornith-1.0-397B	15.6%	estimated ± 8.9 pp, medium confidence
54	Qwen3.7 Max	14.8%	estimated ± 14.0 pp, low confidence
55	MiMo-V2.5	14.5%	estimated ± 14.0 pp, low confidence
56	Gemini 3.5 Flash	14.1%	estimated ± 8.9 pp, medium confidence
57	dots3-note Preview	12.8%	estimated ± 8.9 pp, medium confidence
58	GPT-6 Luna	12.6%	measured
59	Kimi K3	12.6%	measured
60	Claude Opus 4.8	12.2%	estimated ± 8.9 pp, medium confidence
61	Composer 2.5	11.9%	estimated ± 14.0 pp, low confidence
62	Claude Sonnet 4.6	10.6%	estimated ± 14.0 pp, low confidence
63	MiMo-V2.5-Pro	10.6%	estimated ± 14.0 pp, low confidence
64	GLM-5.1	10.2%	estimated ± 14.0 pp, low confidence
65	Seed 2.1 Pro	8.2%	estimated ± 8.9 pp, medium confidence
66	Apodex 1.1	8.0%	estimated ± 8.9 pp, medium confidence
67	GPT-5.4 mini	7.7%	estimated ± 14.0 pp, low confidence
68	Laguna S 2.1	7.4%	estimated ± 8.9 pp, medium confidence
69	Gemini 3 Flash	6.8%	estimated ± 14.0 pp, low confidence
70	Kimi K2.6	6.5%	estimated ± 14.0 pp, low confidence
71	Quasar 438B	6.4%	estimated ± 8.9 pp, medium confidence
72	Qwen3.6 Plus	6.0%	estimated ± 14.0 pp, low confidence
73	Qwen3.8-27B	5.6%	measured
74	Qwen3.7 Plus	5.6%	estimated ± 14.0 pp, low confidence
75	Ornith-1.5-35B-A3B	4.8%	estimated ± 8.9 pp, medium confidence
76	Seed 2.1 Turbo	4.6%	estimated ± 8.9 pp, medium confidence
77	MiniMax M3	2.0%	measured
78	Pokee-Isaac 28B	2.0%	estimated ± 8.9 pp, medium confidence
79	Inkling-Small	1.6%	estimated ± 8.9 pp, medium confidence
80	Ornith-1.0-35B	1.1%	estimated ± 8.9 pp, medium confidence
81	Inkling	1.0%	measured
82	MiniMax M2.7	0.9%	estimated ± 14.0 pp, low confidence
83	Muse Glimmer 30B	0.5%	measured
84	Nemotron 3 Ultra	0.5%	measured
85	A.X K2	0.0%	estimated ± 8.9 pp, low confidence
86	Claude Haiku 4.5	0.0%	estimated ± 14.0 pp, low confidence
87	Command A+	0.0%	estimated ± 14.0 pp, low confidence
88	Gemini 3.1 Flash-Lite	0.0%	estimated ± 14.0 pp, low confidence
89	Gemini 3.5 Flash-Lite	0.0%	estimated ± 8.9 pp, medium confidence
90	GPT-5.4 nano	0.0%	estimated ± 14.0 pp, low confidence
91	Granite 4.2 30B	0.0%	estimated ± 8.9 pp, low confidence
92	Granite 4.2 8B	0.0%	estimated ± 8.9 pp, low confidence
93	Grok 4.20	0.0%	estimated ± 14.0 pp, low confidence
94	Grok 4.3	0.0%	estimated ± 14.0 pp, low confidence
95	K-EXAONE 2.0	0.0%	estimated ± 8.9 pp, low confidence
96	Laguna M.1	0.0%	estimated ± 14.0 pp, low confidence
97	Laguna XS.2	0.0%	estimated ± 14.0 pp, low confidence
98	Ling 3.0 Flash	0.0%	estimated ± 8.9 pp, medium confidence
99	MAI-Code-1.1-Flash	0.0%	estimated ± 8.9 pp, medium confidence
100	Mercury 2.5	0.0%	estimated ± 14.0 pp, low confidence
101	MiniCPM5-2B	0.0%	estimated ± 8.9 pp, low confidence
102	Mistral Medium 3.5 128B	0.0%	estimated ± 14.0 pp, low confidence
103	Nemotron 3.5 Lightning 30B A3B NVFP4	0.0%	estimated ± 8.9 pp, low confidence
104	Ornith-1.0-9B	0.0%	estimated ± 8.9 pp, low confidence
105	Ornith-1.5-9B	0.0%	estimated ± 8.9 pp, low confidence
106	Solar Pro 4	0.0%	estimated ± 8.9 pp, medium confidence
107	Step 3.7 Flash	0.0%	estimated ± 8.9 pp, medium confidence
108	Ternary Bonsai 2 27B	0.0%	estimated ± 8.9 pp, medium confidence

AA Terminal-Bench 4.0 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General