Terminal-Bench

🏆 Leaderboard

As of August 20, 2026, GPT-5.5 is #1 for Terminal-Bench at 82.7%. Ranked by the Terminal-Bench 2.0 score Terminal-Bench leaderboard for 2.0 and 2.1 agent scores, plus API pricing. Rank the best terminal and DevOps LLMs.

Updated August 20, 2026282 models33 providers


GPT-5.5 OpenAI · Proprietary	82.7%	84.7%	76.4%	82.7%	$35.00
Claude Mythos Preview Anthropic · Proprietary	82%	—	—	82%	$60.00
Claude Sonnet 5 Anthropic · Proprietary	80.4%	80.4%	74.5%	80.4%	$12.00
Qwen3.7-Plus Qwen · Proprietary	70.3%	—	52.8%	70.3%	$1.60
Claude Opus 4.8 Anthropic · Proprietary	70.0%	—	71.9%	74.6%	$30.00
Claude Opus 4.7 Anthropic · Proprietary	68.5%	80.2%	68.5%	69.4%	$30.00
MiMo-V2.5-ProOSS Xiaomi · Open Source	68.4%	—	57.3%	68.4%	$1.30
DeepSeek-V4-Pro-MaxOSS DeepSeek · Open Source	67.9%	—	—	67.9%	$5.22
Gemini 3.5 Flash Google · Proprietary	67.4%	76.2%	76.2%	76.2%	$10.50
Gemini 3.1 Pro Google · Proprietary	67.4%	80.2%	70.8%	68.5%	$17.50
MiMo-V2.5OSS Xiaomi · Open Source	65.8%	—	60.7%	65.8%	$0.50
Claude Opus 4.6 Anthropic · Proprietary	65.4%	79.8%	—	65.4%	$30.00
GPT-5.2 Codex OpenAI · Proprietary	64%	66.5%	—	64%	$15.75
GPT-5.4 OpenAI · Proprietary	62.2%	81.8%	—	75.1%	$17.50
Composer 2 Fast Cursor · Proprietary	61.7%	—	—	—	$9.00
Composer 2 Cursor · Proprietary	61.7%	—	—	—	$3.00
GPT-5.4 mini OpenAI · Proprietary	60%	—	54.7%	60%	$5.25
Claude Sonnet 4.6 Anthropic · Proprietary	59.5%	53.4%	57.3%	59.1%	$18.00
Qwen3.7 Max Qwen · Proprietary	59.2%	—	61.0%	69.7%	$5.00
Claude Opus 4.5 Anthropic · Proprietary	58.4%	63.1%	—	59.3%	$30.00
Kimi K2.6OSS Moonshot AI · Open Source	57.3%	—	53.6%	66.7%	$4.93
DeepSeek-V4-Flash-MaxOSS DeepSeek · Open Source	56.9%	—	—	56.9%	$0.42
GPT-5.3 Codex OpenAI · Proprietary	56.7%	78.4%	—	77.3%	$15.75
DeepSeek-V4-Flash-0423OSS DeepSeek · Open Source	56.6%	—	—	56.6%	$0.30
GLM-5OSS Z AI · Open Source	56.2%	52.4%	—	56.2%	$4.20
DeepSeek-V4-Pro-0813OSS DeepSeek · Open Source	56.2%	—	87.9%	—	$5.28
GLM-5.1OSS Z AI · Open Source	53.9%	—	56.9%	69%	$5.80
GPT-5.1 Codex OpenAI · Proprietary	52.8%	60.4%	—	52.8%	$11.25
Qwen3.5-397B-A17BOSS Qwen · Open Source	52.5%	—	—	52.5%	$4.20
GPT-5.2 OpenAI · Proprietary	51.7%	64.9%	—	—	$15.75
Gemini 3 Flash Google · Proprietary	51.7%	64.3%	53.9%	47.6%	$3.50
Qwen3.6-35B-A3BOSS Qwen · Open Source · via OpenRouter	51.5%	24.6%	—	51.5%	$1.14
Step-3.5-FlashOSS StepFun · Open Source	51%	—	—	51%	$0.50
Kimi K2.5OSS Moonshot AI · Open Source	50.8%	43.2%	—	50.8%	$3.68
Qwen3.5-122B-A10BOSS Qwen · Open Source	49.4%	—	—	49.4%	$3.60
MiniMax M2.7OSS MiniMax · Open Source	47.2%	45.1%	48.7%	57%	$1.50
DeepSeek-V3.2OSS DeepSeek · Open Source · via OpenRouter	46.4%	39.6%	—	46.4%	$0.57
GPT-5.4 nano OpenAI · Proprietary	46.3%	—	41.6%	46.3%	$1.45
MiniMax M3OSS MiniMax · Open Source	46.1%	—	66%	—	$1.50
Qwen3.6-27BOSS Qwen · Open Source	44.9%	—	—	59.3%	$4.20
Qwen3.6 Plus Qwen · Proprietary	44.9%	—	53.2%	61.6%	$3.50
GPT-5.1 OpenAI · Proprietary	44.9%	47.6%	—	—	$11.25
Grok 4.3 xAI · Proprietary	43.5%	—	42.0%	—	$3.75
Qwen3.5-27BOSS Qwen · Open Source	41.6%	—	—	41.6%	$2.70
MiniMax M2.5OSS MiniMax · Open Source	41.6%	42.7%	—	—	$1.50
Qwen3.5-35B-A3BOSS Qwen · Open Source	40.5%	—	—	40.5%	$2.25
Gemma 4 31BOSS Google · Open Source	39.3%	—	—	—	$0.54
MiMo-V2-FlashOSS Xiaomi · Open Source	38.5%	30.5%	—	38.5%	$0.40
GLM-4.7OSS Z AI · Open Source	38.2%	33.3%	—	41%	$2.80
Qwen3-Coder 480B A35B InstructOSS Qwen · Open Source · via OpenRouter	37.5%	27.2%	—	37.5%	$2.02

Showing 1–50 of 282 models

Next step

You found the model. Now ship the product.

Auth, billing, and the API layer are already decided. Fork 8 finished AI apps, or have us build the first version with you.

Get the stack — $249 Have us build it with you

The index

All Large Language Models

282 models across 33 providers. Search or jump to a lab — every model page stays linked here.

Cursor

2 models

Composer 2 Composer 2 Fast

xAI

15 models

Grok 4.6 Grok 4.5 Grok 4.3 Grok-4.1 Grok-4.1 Fast Non-Reasoning Grok-4.1 Fast Reasoning Grok-4.1 Thinking Grok 4 Fast Grok Code Fast 1 Grok-4 Fast Non-Reasoning Grok-4 Fast Reasoning Grok-4 Grok-3 Grok-3 Mini Grok-2

DeepSeek

16 models

DeepSeek-V4-Pro-0813 DeepSeek-V4-Flash-0731 DeepSeek-V4-Flash-0423 DeepSeek-V4-Flash-Max DeepSeek-V4-Pro-Max DeepSeek-V3.2 DeepSeek-V3.2 (Non-thinking)DeepSeek-V3.2-Exp DeepSeek-R1-0528 DeepSeek-V3 0324 DeepSeek R1 Distill Llama 70B DeepSeek R1 Distill Qwen 32B DeepSeek-R1 DeepSeek-V3.1 DeepSeek-V3 DeepSeek-V2.5

MiniMax

6 models

MiniMax M3 MiniMax M2.7 MiniMax M2.5 MiniMax M2.1 MiniMax M2 MiniMax M1 80K

AI21 Labs

2 models

Jamba 1.5 Large Jamba 1.5 Mini

Amazon

5 models

Nova 2 Lite Nova 2 Sonic Nova Lite Nova Micro Nova Pro

Baidu

1 models

ERNIE 4.5

ByteDance

2 models

Seed 2.1 Turbo Seed 2.0 Pro

Cohere

3 models

North Mini Code 1.0 Command A+Command R+

IBM

1 models

Granite 3.3 8B Instruct

Inception

1 models

Mercury 2

inclusionAI

1 models

Ling-3.0-flash

LG AI Research

1 models

K-EXAONE-236B-A23B

Meituan

4 models

LongCat-Flash-Lite LongCat-Flash-Thinking-2601 LongCat-Flash-Thinking LongCat-Flash-Chat

Microsoft

5 models

MAI-Code-1.1-Flash Phi 4 Mini Phi-4-multimodal-instruct Phi 4 Phi-3.5-mini-instruct

Moonshot AI

9 models

Kimi K3 Kimi K2.7 Code Kimi K2.7 Code HighSpeed Kimi K2.6 Kimi K2.5 Kimi K2 0905 Kimi K2-Instruct-0905 Kimi K2-Thinking-0905 Kimi K2 Instruct

Nous Research

1 models

Hermes 3 70B

NVIDIA

5 models

Nemotron 3.5 Lightning (30B A3B)Nemotron 3 Ultra (550B A55B)Nemotron 3 Super (120B A12B)Nemotron 3 Nano (30B A3B)Nemotron Nano 9B v2

Poolside

2 models

Laguna S 2.1 Laguna XS 2.1

Sakana AI

1 models

Sakana Namazu

StepFun

1 models

Step-3.5-Flash

Tencent

1 models

Hy3

Thinking Machines

1 models

Inkling-Small

Unisound

1 models

Upstage

1 models

Solar Pro 4

Xiaomi

3 models

MiMo-V2.5-Pro MiMo-V2.5 MiMo-V2-Flash

Which model leads Terminal-Bench right now?

As of August 20, 2026, GPT-5.5 by OpenAI is #1 for Terminal-Bench at 82.7%. Ranked by the Terminal-Bench 2.0 score This board also tracks Terminal-Bench 2.0, Terminal-Bench, Terminal-Bench 2.1, Terminal-Bench 2. Next on the same board: Claude Mythos Preview and Claude Sonnet 5. Related leaders: Kimi K3 on Terminal-Bench at 88.3%; GPT-5.6 Sol on Terminal-Bench 2.1 at 88.8%. This terminal-bench leaderboard ranks models by Terminal-Bench 2.0. Scores come from public evals. Prices are the live API rates in the table above.

Sources: Terminal-Bench (https://www.tbench.ai/); OpenAI API pricing (https://developers.openai.com/api/docs/pricing); Anthropic Claude API pricing (https://platform.claude.com/docs/en/about-claude/pricing); Gemini API pricing (https://ai.google.dev/gemini-api/docs/pricing)

Top 8 for terminal-bench. Ranked by the Terminal-Bench 2.0 score Input and output are dollars per million tokens.
Rank	Model	Terminal-Bench 2.0	Input /M	Output /M
1	GPT-5.5	82.7%	$5.00	$30.00
2	Claude Mythos Preview	82%	$10.00	$50.00
3	Claude Sonnet 5	80.4%	$2.00	$10.00
4	Qwen3.7-Plus	70.3%	$0.32	$1.28
5	Claude Opus 4.8	70.0%	$5.00	$25.00
6	Claude Opus 4.7	68.5%	$5.00	$25.00
7	MiMo-V2.5-Pro	68.4%	$0.43	$0.87
8	DeepSeek-V4-Pro-Max	67.9%	$1.74	$3.48

Related LLM leaderboards

LLM Leaderboard Coding Math Agentic Reasoning Writing Science Finance Long Context Vision Open Source

Individual benchmark pages

Rank one eval at a time. All LLM benchmarks.

HLE SWE Bench Humanity's Last Exam (No Tools)Humanity's Last Exam (With Tools)Terminal-Bench 2.0 Terminal-Bench 2.1 Terminal-Bench 2 Terminal-Bench 3.0 ARC-AGI-1 Verified Chatbot Arena

Terminal-Bench FAQ

Who ranks #1 on the Terminal-Bench leaderboard?

As of August 20, 2026, GPT-5.5 by OpenAI ranks #1 on Terminal-Bench 2.0 at 82.7%. API pricing is $5.00/M input and $30.00/M output.

What are the top models on Terminal-Bench 2.0?

The current Terminal-Bench 2.0 ranking as of August 20, 2026 is 1. GPT-5.5 at 82.7%; 2. Claude Mythos Preview at 82%; 3. Claude Sonnet 5 at 80.4%.

Which terminal-bench model is the cheapest?

DeepSeek-V4-Flash-0423 is the cheapest scored model on this terminal-bench leaderboard at $0.10/M input and $0.20/M output ($0.30 blended). GPT-5.5 still leads Terminal-Bench 2.0 at 82.7%.

Should I always pick the #1 Terminal-Bench 2.0 model?

Not automatically. GPT-5.5 leads Terminal-Bench 2.0, but a cheaper scored model can be the better production choice if the quality gap is small. Use the table to weigh Terminal-Bench 2.0 against input/output price, context window, and related evals.

How often is the Terminal-Bench leaderboard updated?

Scores and API prices on this page are refreshed from published evals and provider rates. The snapshot is labeled August 20, 2026. Treat it as a current index, not a one-off blog post.

What is Terminal-Bench?

Terminal-Bench measures whether an agent can finish realistic shell, sysadmin, and DevOps tasks in a terminal instead of writing a single function.

Should I use Terminal-Bench 2.0 or 2.1?

Use 2.1 when the model has a score. It is the harder current set. We still show 2.0 and the original Terminal-Bench when that is all a lab published.

Is Terminal-Bench the same as SWE-bench?

No. SWE-bench is repository issue fixing. Terminal-Bench is command-line operations. Strong coding models can still fail basic shell workflows.

You found the model. Now ship the product.

All Large Language Models

OpenAI

Anthropic

Google

Cursor

xAI

DeepSeek

MiniMax

Qwen

AI21 Labs

Amazon

Baidu

ByteDance

Cohere

IBM

Inception

inclusionAI

LG AI Research

Meituan

Meta

Microsoft

Mistral

Moonshot AI

Nous Research

NVIDIA

Poolside

Sakana AI

StepFun

Tencent

Thinking Machines

Unisound

Upstage

Xiaomi

Z AI

Which model leads Terminal-Bench right now?

Related LLM leaderboards

Individual benchmark pages

Terminal-Bench FAQ

Who ranks #1 on the Terminal-Bench leaderboard?

What are the top models on Terminal-Bench 2.0?

Which terminal-bench model is the cheapest?

Should I always pick the #1 Terminal-Bench 2.0 model?

How often is the Terminal-Bench leaderboard updated?

What is Terminal-Bench?

Should I use Terminal-Bench 2.0 or 2.1?

Is Terminal-Bench the same as SWE-bench?