PostTrainBench
Rank #20 of 51 · 19 models scored · top score 34.29
unknown
Strongest on contamination resistance, weakest on discrimination. Caveat: the public test set leaves contamination on the table.
Weighted composite. Each signal is computed from the source-backed leaderboard below.
Does it still separate models?
How far from ceiling / clustered at the top?
Public vs held-out; training-leak risk.
Apples-to-apples, or mixed / vendor-optimized?
How old is the benchmark?
| # | Model | Score | Evidence | Measured |
|---|---|---|---|---|
| 1 | GLM-5.2 | 34.29 | unverified | 2026-06-16 |
| 2 | Claude Opus 4.8 | 34.08 | unverified | 2026-05-28 |
| 3 | Claude Opus 4.7 | 28.56 | unverified | 2026-04-16 |
| 4 | GPT-5.5 | 25.02 | unverified | 2026-04-23 |
| 5 | Claude Opus 4.6 | 24.82 | unverified | 2026-02-05 |
| 6 | Gemini 3.1 Pro | 21.59 | unverified | 2026-02-19 |
| 7 | GPT-5.2 | 21.38 | unverified | 2025-12-11 |
| 8 | GPT-5.4 | 20.23 | unverified | 2026-03-05 |
| 9 | Gemini 3 Pro | 18.12 | unverified | 2025-11-18 |
| 10 | GPT-5.3 Codex | 17.76 | unverified | 2026-02-05 |
| 11 | Claude Opus 4.5 | 17.29 | unverified | 2025-11-24 |
| 12 | Claude Sonnet 4.6 | 16.42 | unverified | 2026-02-17 |
| 13 | GLM-5 | 13.88 | unverified | 2026-02-11 |
| 14 | Kimi K2.5 | 10.26 | unverified | 2026-02-02 |
| 15 | Claude Sonnet 4.5 | 9.94 | unverified | 2025-09-29 |
| 16 | MiniMax-M2.5 | 9.5 | unverified | 2026-02-12 |
| 17 | GLM-4.7 | 7.48 | unverified | 2025-12-22 |
| 18 | Qwen3-Max | 7.42 | unverified | 2025-09-05 |
| 19 | Kimi K2 Thinking | 7.25 | unverified | 2025-11-06 |