TerminalBench v2.1 benchmark

116 AI models ranked on TerminalBench v2.1. Each row shows the model’s score next to its cheapest API price per 1M tokens where it has one, so you can weigh quality against cost.

Top model today

GPT-5.6 Sol (OpenAI) leads at 90%, from $5.00 per 1M input.

Data via GitHub: harbor-framework/terminal-bench-2-1

Models ranked

116

on this benchmark

Top score

90%

current leader

Type

Single eval

benchmark

Updated

15 July 2026

last source fetch

Cost vs TerminalBench v2.1

72 priced models
Most attractive quadrant (cheap + high)
= reasoning model (extended thinking)

Each labelled name links to that model’s page. Curated to the cost-efficiency frontier, top scorers and cheapest so labels stay readable; the full field is in the leaderboard table below (and via “select all”). Reasoning models (extended “thinking” before answering) carry a ring around the dot.

Models ranked on TerminalBench v2.1

highest score first · 116 models
Maker
#
1GPT-5.6 SolOpenAI90%xhigh$5.00$30.00
2GPT-5.6 TerraOpenAI88%$2.50$15.00
3Claude Opus 4.8Anthropic85%$5.00$25.00
4Claude Fable 5Anthropic85%$10.00$50.00
5GPT-5.5OpenAI84%$5.00$30.00
6Claude Opus 4.7Anthropic83%$5.00$25.00
7Grok 4.5xAI82%$2.00$6.00
8GPT-5.6 LunaOpenAI81%$1.00$6.00
9Claude Sonnet 5Anthropic81%$2.00$10.00
10Gemini 3.5 FlashGoogle79%$1.50$9.00
11GPT-5.4OpenAI78%$2.50$15.00
12Muse Spark 1.1Other78%$1.25$4.25
13GLM 5.2Zhipu78%$1.40$4.40
14Qwen3.7 MaxAlibaba75%n/an/a
15Gemini 3.1 Pro PreviewGoogle74%$2.00$12.00
16Claude Sonnet 4.6Anthropic71%reasoning: adaptive$3.00$15.00
17KAT-Coder-Pro V2Other70%$0.30$1.20
18Nex-N2-ProNex-agi68%n/an/a
19Kimi K2.7 CodeMoonshot67%$0.95$4.00
20Kimi K2.6Moonshot66%n/an/a
21MiniMax M3MiniMax65%$0.30$1.20
22MiMo-V2.5-ProXiaomi65%$1.00$3.00
23Hy3-previewTencent64%n/an/a
24DeepSeek V4 ProDeepSeek64%$0.43$0.87
25MiMo-V2.5Xiaomi64%n/an/a
26Muse SparkMeta62%n/an/a
27Mimo V2 FlashXiaomi62%$0.10$0.30
28GLM 5.1Zhipu62%$1.05$3.50
29DeepSeek V4 FlashDeepSeek62%n/an/a
30Qwen3.6 PlusAlibaba61%$0.33$1.95
31Qwen3.7 PlusAlibaba61%n/an/a
32GPT-5.4 NanoOpenAI61%$0.20$1.25
33Qwen3.6 27BAlibaba61%$0.15$0.50
34JT-4.1 Flash 236B A21BChina-mobile60%n/an/a
35GPT-5.4 MiniOpenAI59%$0.75$4.50
36Claude 4.5 SonnetAnthropic56%reasoning: true$3.00$15.00
37MiniMax M2.7MiniMax55%$0.26$0.55
38Nemotron 3 UltraNVIDIA54%n/an/a
39GPT-5.1OpenAI52%$1.25$10.00
40Grok Build 0.1 0616Spacexai52%n/an/a
41Qwen3.5 397B A17BAlibaba51%$0.60$3.60
42Mistral Medium 3.5Mistral51%$1.50$7.50
43LongCat 2.0Longcat50%n/an/a
44GLM 4.6Zhipu49%reasoning: true$0.40$1.75
45Qwen3.5-122B-A10BAlibaba48%$0.25$1.75
46DeepSeek V3.2DeepSeek47%reasoning: true$0.27$0.40
47Kimi K2.5Moonshot46%$0.50$2.80
48GLM 4.7Zhipu45%$0.40$1.50
49Qwen3.6 35B A3BAlibaba45%$0.14$0.45
50DeepSeek V3.1 TerminusDeepSeek45%reasoning: true$0.27$1.00
150 of 116
Page 1 of 3
Prices are on-demand list rates. Enterprise commitments can be materially lower. Cost per task uses best-effort token usage.

Frequently asked questions

What is the TerminalBench v2.1 benchmark?

A verified release of the Terminal-Bench agentic benchmark: 89 curated real-world tasks (model training, sysadmin, data processing, security, software engineering) an agent must complete in a sandboxed container terminal. It refines v2.0 for bugs, timeouts and reward-hacking robustness, scored by automated verification tests.

Which AI model scores highest on TerminalBench v2.1?

GPT-5.6 Sol (OpenAI) leads with 90%, from $5.00 per 1M input tokens.

How many models are ranked on TerminalBench v2.1?

116 models carry a TerminalBench v2.1 score. Scores come from Artificial Analysis; where a model has a live API price it is verified against LiteLLM and OpenRouter, and models with no current price are listed without one.

Other benchmarks

compare the same models on a different eval
Benchmark scores via Artificial Analysis · pricing verified against LiteLLM + OpenRouter · refreshed daily · updated 15 July 2026

Every weekday

AI moves fast. Here's your debrief.

News, analysis, tools, and more.

For people who build with AI