Skip to content
agents benchmark · included in ECI

APEX-Agents

APEX-Agents benchmark (score column: Pass@1 score).
Results
34
Random baseline
0.0%
Score ceiling
100%
Released
21 Jan 2026

Score by model release date

Best score by organisation

#ModelScoreRelativeEvidence
1Claude Opus 5.5
Anthropic
73.5%Source ↗
2Claude Fable 5.1 (unknown)
Anthropic
68.6%Source ↗
3Gemini 3.7 Flash (unknown)
Google DeepMind
67.8%Source ↗
4Claude Opus 5
Anthropic
65.8%Source ↗
5Grok 4.6 (unknown)
xAI
65.3%Source ↗
6GPT-6 Astra (unknown thinking)
OpenAI
64.7%Source ↗
7Gemini 3.8 Flash (unknown)
Google DeepMind
64.3%Source ↗
8Claude Fable 5
Anthropic
63.6%Source ↗
9Claude Fable 5.1
Anthropic
59.7%Source ↗
10GPT-5.6 Terra
OpenAI
58.2%Source ↗
11Muse Spark 1.3 (unknown)
Meta AI
57.8%Source ↗
12GLM-5.3 (unknown thinking)
Z.ai (Zhipu AI)
56.6%Source ↗
13Grok 4.5 (unknown thinking)
xAI
56.2%Source ↗
14GPT-5.5 (unknown thinking)
OpenAI
55.1%Source ↗
15Claude Sonnet 5 (unknown thinking)
Anthropic
54.5%Source ↗
16GLM-5.3-Flash (unknown) Open source
Z.ai (Zhipu AI)
52.8%Source ↗
17GPT-5.4 (unknown thinking)
OpenAI
52.4%Source ↗
18GPT-5.6 Sol (pro, max)
OpenAI
51.4%Source ↗
19Kimi K3 (unknown) Open source
Moonshot
50.6%Source ↗
20Claude Opus 4.7
Anthropic
49.2%Source ↗
21Claude Opus 4.8
Anthropic
48.9%Source ↗
22DeepSeek V4 Pro 0813 (unknown) Open source
DeepSeek
47.3%Source ↗
23Gemini 3.6 Flash
Google
46.9%Source ↗
24Claude Opus 4.6
Anthropic
46.3%Source ↗
25Claude Sonnet 4.6
Anthropic
43.0%Source ↗
26GLM 5.1 Open source
Z.ai
40.9%Source ↗
27MiniMax M3 Open source
MiniMax
37.7%Source ↗
28Kimi K2.7 Code Open source
MoonshotAI
37.6%Source ↗
29Muse Spark 1.2
Meta
36.4%Source ↗
30Gemini 3.1 Pro Preview
Google
35.3%Source ↗
31Inkling Open source
Thinking Machines
33.8%Source ↗
32Gemini 3.5 Flash
Google
27.5%Source ↗
33Qwen3.5 397B A17B Open source
Qwen
24.9%Source ↗
34gpt-oss-120b Open source
OpenAI
4.4%Source ↗

Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.