Posttrainbench

Models scored

30

Best score

0.34

Metric

Average (%)

Posttrainbench is an AI evaluation tracked by Epoch AI, with 30 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.

Scored models span September 2025 to June 2026. The best score recorded is 0.34, measured as Average (%). Models come from Z.ai (Zhipu AI), Anthropic, OpenAI, Google DeepMind, Moonshot, MiniMax, Alibaba.

Full record

Score metric: Average (%)
Models scored: 30
Best score recorded: 0.34
Earliest model scored: Sep 24, 2025
Most recent model scored: Jun 16, 2026
Third-party reported: Yes

Leaderboard

01	glm-5.2_max	Z.ai (Zhipu AI)	0.34
02	claude-opus-4-8_max	Anthropic	0.34
03	claude-opus-4-8_high	Anthropic	0.34
04	claude-opus-4-7_xhigh	Anthropic	0.29
05	gpt-5.5_xhigh	OpenAI	0.25
06	claude-opus-4-6_unknown	Anthropic	0.25
07	claude-opus-4-6	Anthropic	0.23
08	claude-opus-4-6_high	Anthropic	0.23
09	gemini-3.1-pro-preview	Google DeepMind	0.22
10	gpt-5.2-2025-12-11_medium	OpenAI	0.21
11	gpt-5.2-2025-12-11_none	OpenAI	0.21
12	gpt-5.4-2026-03-05_high	OpenAI	0.2
13	gpt-5.1-codex-max	OpenAI	0.2
14	gemini-3-pro-preview	Google DeepMind	0.18
15	gpt-5.3-codex_high	OpenAI	0.18
16	claude-opus-4-5-20251101	Anthropic	0.17
17	gpt-5.2-codex	OpenAI	0.17
18	claude-sonnet-4-6	Anthropic	0.16
19	glm-5	Z.ai (Zhipu AI)	0.14
20	gpt-5.3-codex	OpenAI	0.14
21	kimi-k2.5	Moonshot	0.1
22	claude-sonnet-4-5-20250929	Anthropic	0.1
23	MiniMax-M2.5	MiniMax	0.1
24	MiniMax-M2.1	MiniMax	0.09
25	glm-4.7	Z.ai (Zhipu AI)	0.07
26	qwen3-max-2025-09-23	Alibaba	0.07
27	kimi-k2-thinking	Moonshot	0.07

More benchmarks

Epoch Capabilities Index MMLU Terminal-Bench GPQA Diamond OTIS Mock AIME 2024-2025 GSM8K ARC-AGI ARC-AGI-2

SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.

← All benchmarks