Agentic · terminal

AA Terminal-Bench 2.1 leaderboard

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

As of 2026-10-07, the highest measured score on AA Terminal-Bench 2.1 is 91.4% by Claude Fable 5.1. 92 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	91.4%	measured
2	Claude Opus 5	88.8%	estimated ± 4.4 pp, high confidence
3	GPT-6 Astra	88.4%	measured
4	Gemini 3.8 Flash	87.6%	measured
5	Grok 4.6	87.6%	estimated ± 4.4 pp, high confidence
6	GPT-5.5	87.1%	estimated ± 4.4 pp, high confidence
7	Claude Sonnet 5.5	86.8%	estimated ± 2.6 pp, medium confidence
8	Claude Opus 5.5	86.8%	estimated ± 2.6 pp, medium confidence
9	Gemini 4 Argon	86.8%	estimated ± 2.6 pp, high confidence
10	GPT-6.1 Sol	86.8%	estimated ± 2.6 pp, high confidence
11	GPT-6 Sol	86.8%	estimated ± 2.6 pp, high confidence
12	MiMo-V2.6-Pro	86.7%	estimated ± 2.6 pp, high confidence
13	Step 5 Preview	86.6%	estimated ± 2.6 pp, high confidence
14	Claude Haiku 5.5	86.6%	estimated ± 2.6 pp, high confidence
15	DeepSeek V4.1 Flash	86.5%	estimated ± 2.6 pp, high confidence
16	Mistral Large 4	86.5%	estimated ± 2.6 pp, high confidence
17	Grok 4.7	86.4%	estimated ± 2.6 pp, high confidence
18	Gemini 3.6 Flash	86.2%	estimated ± 4.4 pp, high confidence
19	GPT-6 Luna	85.1%	estimated ± 2.6 pp, high confidence
20	Gemini 3.1 Pro	85.0%	estimated ± 4.4 pp, high confidence
21	Kimi K3	85.0%	measured
22	SWE-2	84.9%	estimated ± 3.5 pp, medium confidence
23	GPT-5.6 Sol	84.9%	estimated ± 3.5 pp, medium confidence
24	Claude Mythos 5	84.9%	estimated ± 3.5 pp, high confidence
25	DeepSeek V4 Pro 0813	84.9%	estimated ± 3.5 pp, high confidence
26	MiMo-V2.6-Flash	84.9%	estimated ± 3.5 pp, high confidence
27	GPT-5.6 Terra	84.9%	estimated ± 3.5 pp, high confidence
28	Qwen3.8 Max	84.9%	estimated ± 3.5 pp, high confidence
29	Ornith-1.5-397B	84.9%	estimated ± 3.5 pp, high confidence
30	Gemini 3.7 Flash	84.9%	estimated ± 3.5 pp, high confidence
31	Hy4 preview	84.9%	estimated ± 3.5 pp, high confidence
32	GPT-5.6 Luna	84.9%	estimated ± 3.5 pp, high confidence
33	Claude Fable 5	84.9%	estimated ± 3.5 pp, high confidence
34	Grok 4.5	84.8%	estimated ± 3.5 pp, high confidence
35	Muse Spark 1.2	84.8%	estimated ± 3.5 pp, high confidence
36	DeepSeek V4 Flash 0731	84.8%	estimated ± 3.5 pp, high confidence
37	Sakana Fugu-Ultra	84.8%	estimated ± 3.5 pp, high confidence
38	Ember-1	84.8%	estimated ± 3.5 pp, high confidence
39	SWE-1.7	84.7%	estimated ± 3.5 pp, high confidence
40	GLM-5.2	84.7%	estimated ± 3.5 pp, high confidence
41	Claude Sonnet 5	84.6%	estimated ± 3.5 pp, high confidence
42	Sakana Fugu	84.6%	estimated ± 3.5 pp, high confidence
43	Beam	84.6%	estimated ± 3.5 pp, high confidence
44	Muse Spark 1.1	84.6%	estimated ± 3.5 pp, high confidence
45	GLM-5.3-Flash	84.3%	measured
46	Muse Spark 1.3	84.3%	measured
47	Atria Dawn Preview	84.3%	estimated ± 3.5 pp, high confidence
48	Ornith-1.0-397B	84.1%	estimated ± 3.5 pp, high confidence
49	GLM-5.3	83.9%	measured
50	Claude Opus 4.7	83.8%	estimated ± 4.4 pp, high confidence
51	Gemini 3.5 Flash	83.5%	estimated ± 3.5 pp, high confidence
52	Kimi K2.7 Code	82.9%	estimated ± 4.4 pp, high confidence
53	dots3-note Preview	82.9%	estimated ± 3.5 pp, high confidence
54	Claude Opus 4.8	82.5%	estimated ± 3.5 pp, high confidence
55	Qwen3.8-27B	79.8%	measured
56	Qwen3.7 Max	77.8%	estimated ± 4.4 pp, high confidence
57	MiMo-V2.5	77.5%	estimated ± 4.4 pp, high confidence
58	Seed 2.1 Pro	77.2%	estimated ± 3.5 pp, high confidence
59	Apodex 1.1	76.7%	estimated ± 3.5 pp, high confidence
60	Laguna S 2.1	75.2%	estimated ± 3.5 pp, high confidence
61	Composer 2.5	74.6%	estimated ± 4.4 pp, high confidence
62	Claude Sonnet 4.6	73.0%	estimated ± 4.4 pp, high confidence
63	MiMo-V2.5-Pro	73.0%	estimated ± 4.4 pp, high confidence
64	Quasar 438B	72.8%	estimated ± 3.5 pp, high confidence
65	GLM-5.1	72.4%	estimated ± 4.4 pp, high confidence
66	GPT-5.4 mini	68.7%	estimated ± 4.4 pp, high confidence
67	Ornith-1.5-35B-A3B	68.3%	estimated ± 3.5 pp, high confidence
68	Seed 2.1 Turbo	67.6%	estimated ± 3.5 pp, high confidence
69	Gemini 3 Flash	67.2%	estimated ± 4.4 pp, high confidence
70	Kimi K2.6	66.6%	estimated ± 4.4 pp, high confidence
71	Qwen3.6 Plus	65.8%	estimated ± 4.4 pp, high confidence
72	MiniMax M3	65.2%	measured
73	Qwen3.7 Plus	64.9%	estimated ± 4.4 pp, high confidence
74	Pokee-Isaac 28B	60.6%	estimated ± 3.5 pp, high confidence
75	Inkling-Small	59.7%	estimated ± 3.5 pp, high confidence
76	Ornith-1.0-35B	58.7%	estimated ± 3.5 pp, high confidence
77	MAI-Code-1.1-Flash	56.5%	estimated ± 3.5 pp, high confidence
78	MiniMax M2.7	55.1%	estimated ± 4.4 pp, high confidence
79	Inkling	55.1%	measured
80	Nemotron 3 Ultra	53.9%	measured
81	Step 3.7 Flash	53.4%	estimated ± 3.5 pp, high confidence
82	Ling 3.0 Flash	52.6%	estimated ± 3.5 pp, high confidence
83	Solar Pro 4	52.6%	estimated ± 3.5 pp, high confidence
84	Gemini 3.5 Flash-Lite	52.3%	estimated ± 3.5 pp, high confidence
85	Ternary Bonsai 2 27B	52.2%	estimated ± 3.5 pp, high confidence
86	Ornith-1.5-9B	52.1%	estimated ± 3.5 pp, medium confidence
87	K-EXAONE 2.0	52.1%	estimated ± 3.5 pp, medium confidence
88	Ornith-1.0-9B	52.1%	estimated ± 3.5 pp, medium confidence
89	A.X K2	52.1%	estimated ± 3.5 pp, medium confidence
90	Granite 4.2 30B	52.1%	estimated ± 3.5 pp, medium confidence
91	Granite 4.2 8B	52.1%	estimated ± 3.5 pp, medium confidence
92	MiniCPM5-2B	52.1%	estimated ± 3.5 pp, medium confidence
93	Nemotron 3.5 Lightning 30B A3B NVFP4	52.1%	estimated ± 3.5 pp, medium confidence
94	Muse Glimmer 30B	51.7%	measured
95	Grok 4.20	42.1%	estimated ± 4.4 pp, medium confidence
96	Claude Haiku 4.5	40.8%	estimated ± 4.4 pp, medium confidence
97	Grok 4.3	35.0%	estimated ± 4.4 pp, medium confidence
98	GPT-5.4 nano	34.0%	estimated ± 4.4 pp, medium confidence
99	Mistral Medium 3.5 128B	26.3%	estimated ± 4.4 pp, medium confidence
100	Mercury 2.5	14.8%	estimated ± 4.4 pp, medium confidence
101	Gemini 3.1 Flash-Lite	14.0%	estimated ± 4.4 pp, medium confidence
102	Laguna M.1	14.0%	estimated ± 4.4 pp, medium confidence
103	Laguna XS.2	3.0%	estimated ± 4.4 pp, medium confidence
104	Command A+	0.3%	estimated ± 4.4 pp, medium confidence

AA Terminal-Bench 2.1 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General