Terminal-Bench 2.1

A verified refresh of Terminal-Bench 2.0: 89 curated tasks across software engineering, system administration, data processing, model training, and security, with environment and instruction fixes so scores reflect agent capability rather than environment gaps.

Models: 114 current, 243 in total
Median: 38.6%
Range: 0.0%–89.9%

Score distribution

Lowest: 0.0%
Median: 38.6%
Highest: 89.9%

Leaderboard

Include deprecated models

Rank	Model	Score
1	GPT-6 Astra (High) OpenAI	89.9%
2	GPT-6 Astra (Medium) OpenAI	89.5%
3	GPT-6 Astra (Xhigh) OpenAI	89.1%
4	Qwen3.8 Max (0902) Alibaba	88.8%
5	GPT-6 Astra (Max) OpenAI	88.4%
6	GPT-6 Astra (Low) OpenAI	88.0%
7	Gemini 3.8 Flash (High) Google	87.6%
8	Qwen3.8-Flash-Next Alibaba	86.1%
9	GLM-5.3-Flash Z AI	84.3%
10	Gemini 3.8 Flash (Medium) Google	83.9%
11	GLM-5.3 (Max) Z AI	83.9%
12	Gemini 3.8 Flash (Low) Google	83.1%
13	Qwen3.8 2.4T A95B Alibaba	82.0%
14	Inkling 2 Small Thinking Machines	81.3%
15	Qwen3.8 27B (Xhigh) Alibaba	79.8%
16	DeepSeek V4 Pro 0813 (Max) DeepSeek	78.7%
17	Muse Spark Open (Xhigh) Meta	76.4%
18	Motif 3 Motif Technologies	74.9%
19	DeepSeek V4 Flash Vision (Max) DeepSeek	74.2%
20	Gemini 3.1 Pro Preview Google	73.8%
21	K2 Horizon 375B A23B Institute of Foundation Models	71.9%
22	KAT Coder Pro V2 KwaiKAT	70.0%
23	Agnes 2.5 Pro Beta Sapiens AI	69.7%
24	Apodex 1.1 Apodex	69.7%
25	Quasar 438B (Max, Based on GLM-5.2) Multiverse Computing	69.3%
26	Nex-N2-Pro (Based on Qwen3.5-397B-A17B) Nex AGI	67.8%
27	Kimi K2.7 Code Kimi	67.4%
28	Qwen3.8 27B (Low) Alibaba	67.4%
29	MiniMax-M3 MiniMax	65.2%
30	Qwen3.8 27B (Medium) Alibaba	65.2%
31	Hy3 Tencent	64.4%
32	Ling-3.0-flash-VL InclusionAI	64.4%
33	Ling-3.0-flash-Fin InclusionAI	62.5%
34	Qwen3.7 Plus Alibaba	61.0%
35	JT-4.1 Flash 236B A21B (Non-reasoning) China Mobile	59.6%
36	Solar Pro 4 Upstage	57.3%
37	K2 Horizon MoVA 36B A4B Institute of Foundation Models	56.6%
38	Inkling (Xhigh) Thinking Machines	55.1%
39	Inkling Small Thinking Machines	55.1%
40	Nemotron 3 Ultra 550B A55B (Reasoning) NVIDIA	53.9%
41	Gemini 3.5 Flash-Lite Google	53.6%
42	Muse Glimmer (High) Meta	51.7%
43	Qwen3.5 397B A17B (Reasoning) Alibaba	51.3%
44	Mistral Medium 3.5 Mistral	50.6%
45	LongCat 2.0 LongCat	50.2%
46	Qwen3.8 27B (Non-reasoning) Alibaba	49.1%
47	Qwen3.5 122B A10B (Reasoning) Alibaba	47.6%
48	Qwen3.5 122B A10B (Non-reasoning) Alibaba	47.2%
49	Qwen3.6 35B A3B (Reasoning) Alibaba	44.9%
50	K2 Horizon 7B Institute of Foundation Models	44.2%

1
GPT-6 Astra (High)
OpenAI
89.9%
2
GPT-6 Astra (Medium)

Data from Artificial Analysis

Command Palette

Terminal-Bench 2.1

Score distribution

Leaderboard