Skip to content

Benchmark

PostTrainBench

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

Models measured
27
Scores recorded
35
Sources
3
IQ weight
1.58

Score distribution

33.7%
Mean
36.5%
Median
8%
Std dev
22.6%
10th percentile
43.8%
90th percentile
16.5%
Min
16.5%45.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.58 for chat models, measured over 27 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 4 Argon· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim45.3%
2Claude Fable 5.1max· Epoch AI Benchmarking Hub data44.7%
3Claude Fable 5.2max· Epoch AI Benchmarking Hub data44.6%
4Claude Opus 5.5max· Epoch AI Benchmarking Hub data43.8%
5GPT-6 Astramax· Epoch AI Benchmarking Hub data41.9%
6Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data41%
7Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data40.4%
8GLM 5.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim39.8%
9GLM 5.3Claude Code 2.1.207 · max · 3-run weighted average · documented H100 fallback· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.39.8%
10MiniMax M3.1 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim39.3%
11GPT-6.1 Solmax· Epoch AI Benchmarking Hub data38.8%
12Claude Fable 5max· Epoch AI Benchmarking Hub data37.3%
13MiniMax M3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim37.1%
14MiniMax M3Claude Code · Ralph-Loop · 12h· Reported by MiniMax Research· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.37.1%
15GLM 5.3 Flashmax· Epoch AI Benchmarking Hub data36.8%
16Kimi K3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
36.6%
17Kimi K3official Harbor · H100 · max effort · temp 1.0 · top_p 1.0· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.36.6%
18Kimi K3.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim36.5%
19Claude Opus 5· Epoch AI Benchmarking Hub data35.7%
20Hy4 preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim35.6%
21GLM 5.3max· Epoch AI Benchmarking Hub data34.5%
22GLM 5.2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim34.3%
23GLM 5.21M context · max effort · 128K output· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.34.3%
24GPT-5.6 Solmax· Epoch AI Benchmarking Hub data31%
25Claude Opus 4.8max· Epoch AI Benchmarking Hub data31%
26Claude Opus 4.8high· Epoch AI Benchmarking Hub data30.7%
27GLM 5.2max· Epoch AI Benchmarking Hub data30.4%
28Claude Opus 4.7xhigh· Epoch AI Benchmarking Hub data24.3%
29Muse Wonder 1.0xhigh· Epoch AI Benchmarking Hub data24.2%
30GPT 5.5xhigh· Epoch AI Benchmarking Hub data23.9%
31Grok 4.5high· Epoch AI Benchmarking Hub data22.6%
32Doubao Seed 2.1 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim18.3%
33Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data18.1%
34GPT 5.4high· Epoch AI Benchmarking Hub data17.8%
35Doubao Seed 2.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim16.5%

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.