Skip to content

Benchmark

Terminal-Bench

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.

Models measured
61
Scores recorded
67
Sources
3
IQ weight
1.73

Score distribution

41.8%
Mean
39.5%
Median
18.5%
Std dev
21.8%
10th percentile
66.5%
90th percentile
3.4%
Min
3.4%84.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.73 for chat models, measured over 61 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT 5.5· Epoch AI Benchmarking Hub data84.7%
2GPT 5.4· Epoch AI Benchmarking Hub data81.8%
3Claude Opus 4.7· Epoch AI Benchmarking Hub data80.2%
4Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data80.2%
5Claude Opus 4.6· Epoch AI Benchmarking Hub data79.8%
6GPT-5.3 Codex· Epoch AI Benchmarking Hub data78.4%
7Gemini 3 Pro Preview· Epoch AI Benchmarking Hub data69.4%
8GPT-5.2 Codex· Epoch AI Benchmarking Hub data66.5%
9GPT 5.2· Epoch AI Benchmarking Hub data64.9%
10GPT 5.2medium· Epoch AI Benchmarking Hub data64.9%
11Gemini 3 Flash· Epoch AI Benchmarking Hub data64.3%
12Claude Opus 4.5· Epoch AI Benchmarking Hub data63.1%
13GPT-5.1 Codex Mini· Epoch AI Benchmarking Hub data61.6%
14GPT-5.1 Codex Max· Epoch AI Benchmarking Hub data60.4%
15Claude Opus 4.5128k· Epoch AI Benchmarking Hub data59.1%
16GPT-5.1 Codex· Epoch AI Benchmarking Hub data57.8%
17Grok 4.20· Epoch AI Benchmarking Hub data57.3%
18Claude Sonnet 4.6· Epoch AI Benchmarking Hub data53.4%
19GLM 5· Epoch AI Benchmarking Hub data52.4%
20Claude Sonnet 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
50%
21GPT 5· Epoch AI Benchmarking Hub data49.6%
22GPT 5medium· Epoch AI Benchmarking Hub data49.6%
23MiniMax M2.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
47.9%
24GPT 5.1medium· Epoch AI Benchmarking Hub data47.6%
25GPT 5.1· Epoch AI Benchmarking Hub data47.6%
26Kimi K2 Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
47.1%
27MiniMax M2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.
Values from 2 sources
46.3%
28MiniMax M2.7· Epoch AI Benchmarking Hub data45.1%
29GPT-5 Codex· Epoch AI Benchmarking Hub data44.3%
30Claude Opus 4.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
43.3%
31Kimi K2.5· Epoch AI Benchmarking Hub data43.2%
32MiniMax M2.5· Epoch AI Benchmarking Hub data42.7%
33GLM 5.3· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim41.2%
34Claude Haiku 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
41%
35GLM 4.6· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
40.5%
36DeepSeek V3.2· Epoch AI Benchmarking Hub data39.6%
37LongCat Flash Chat· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim39.5%
38Claude Opus 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim39.2%
39DeepSeek V3.2 Exp· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim37.7%
40GLM 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim37.5%
41Claude Sonnet 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim35.5%
42GLM 5.3 Flash· Hugging Face Official Benchmarks data· Reported by the model's developer. The source relaying it did not verify it. claim35.5%
43Claude Sonnet 3.7· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim35.2%
44GPT-5 mini· Epoch AI Benchmarking Hub data34.8%
45longcat-flash-lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim33.8%
46GLM 4.7· Epoch AI Benchmarking Hub data
Values from 2 sources
33.4%
47Gemini 2.5 Pro· Epoch AI Benchmarking Hub data32.6%
48Nova 2 Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim32.5%
49GPT-5 minimedium· Epoch AI Benchmarking Hub data31.9%
50DeepSeek V3.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim31.3%

Showing 1–50 of 67 results.

Data credits: Data by LLM Stats · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Hugging Face official benchmark leaderboards; licence varies by benchmark