Skip to content

Benchmark

Apex Agents

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

Models measured
72
Scores recorded
83
Sources
3
IQ weight
2.12

Score distribution

50.7%
Mean
51.6%
Median
16.9%
Std dev
27.9%
10th percentile
70%
90th percentile
4.4%
Min
4.4%84.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.12 for chat models, measured over 72 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 Carbon· Epoch AI Benchmarking Hub data84.2%
2Gemini 4 Neonxhigh· Epoch AI Benchmarking Hub data82.9%
3Gemini 4 Neonhigh· Epoch AI Benchmarking Hub data82.7%
4Gemini 4 Argon· Epoch AI Benchmarking Hub data82.4%
5Claude Fable 5.2· Epoch AI Benchmarking Hub data80.1%
6Gemini 4 Hydrogen· Epoch AI Benchmarking Hub data78%
7Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data75.5%
8Claude Opus 5.5max· Epoch AI Benchmarking Hub data73.5%
9GPT-6.2 Astramax· Epoch AI Benchmarking Hub data72.2%
10Grok 4.7· Epoch AI Benchmarking Hub data70%
11DeepSeek V4.1 Flash 1017max· Epoch AI Benchmarking Hub data69.8%
12Claude Fable 5.1· Epoch AI Benchmarking Hub data68.6%
13GPT-6.1 Terramax· Epoch AI Benchmarking Hub data68.3%
14Gemini 3.7 Flash· Epoch AI Benchmarking Hub data67.8%
15Kimi K3.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
67.2%
16GPT-6.2 Solmax· Epoch AI Benchmarking Hub data66.6%
17Muse Wonder 1.0max· Epoch AI Benchmarking Hub data66.1%
18Claude Opus 5max· Epoch AI Benchmarking Hub data65.8%
19Grok 4.6· Epoch AI Benchmarking Hub data
Values from 2 sources
65.3%
20GPT-6 Astra· Epoch AI Benchmarking Hub data64.7%
21Gemini 3.8 Flash· Epoch AI Benchmarking Hub data64.3%
22Muse Ember 1.0· Epoch AI Benchmarking Hub data64.3%
23ERNIE 5.5 Flash Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim64.2%
24Claude Fable 5· Epoch AI Benchmarking Hub data63.6%
25Qwen3.8 Max· Epoch AI Benchmarking Hub data63.3%
26Hy4 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim63.1%
27GPT-6.1 Sol· LLM Stats data60%
28GPT-6.1 Solmax· Epoch AI Benchmarking Hub data60%
29Claude Fable 5.1high· Epoch AI Benchmarking Hub data59.7%
30Muse Spark 1.3· Epoch AI Benchmarking Hub data58.6%
31GPT-5.6 Terramax· Epoch AI Benchmarking Hub data58.2%
32GLM 5.3· Epoch AI Benchmarking Hub data56.6%
33Grok 4.5· Epoch AI Benchmarking Hub data56.2%
34Claude Haiku 5.5max· Epoch AI Benchmarking Hub data56.1%
35GPT 5.5· Epoch AI Benchmarking Hub data55.1%
36Claude Sonnet 5· Epoch AI Benchmarking Hub data54.5%
37MiMo V2.6 Pro· Epoch AI Benchmarking Hub data54.4%
38GPT-6 Solmax· Epoch AI Benchmarking Hub data54.3%
39GLM 5.3 Flash· Epoch AI Benchmarking Hub data52.8%
40Claude Opus 5.5medium· Epoch AI Benchmarking Hub data52.5%
41GPT 5.4· Epoch AI Benchmarking Hub data52.4%
42Mistral Large 4high· Epoch AI Benchmarking Hub data51.6%
43GPT-5.6 Solpro / max· Epoch AI Benchmarking Hub data51.4%
44Claude Opus 4.7max· Epoch AI Benchmarking Hub data49.2%
45Claude Opus 4.8max· Epoch AI Benchmarking Hub data48.9%
46Kimi K3· Epoch AI Benchmarking Hub data
Values from 2 sources
47.8%
47Muse Spark 1.3max· Epoch AI Benchmarking Hub data47.6%
48Qwen 3.8 27B· Epoch AI Benchmarking Hub data47.5%
49DeepSeek V4 Pro 0813· Epoch AI Benchmarking Hub data47.3%
50Gemini 3.6 Flash· Epoch AI Benchmarking Hub data46.9%

Showing 1–50 of 83 results.

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.