GSO-Bench
Rank #16 of 51 · 22 models scored · top score 44.12
unknown
Strongest on discrimination, weakest on contamination resistance. 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 | Claude Opus 4.7 | 44.12 | unverified | 2026-04-16 |
| 2 | Claude Opus 4.6 | 41.2 | unverified | 2026-02-05 |
| 3 | GPT-5.5 | 40.2 | unverified | 2026-04-23 |
| 4 | GPT-5.4 | 31.37 | unverified | 2026-03-05 |
| 5 | GPT-5.2 | 27.4 | unverified | 2025-12-11 |
| 6 | Claude Opus 4.5 | 26.5 | unverified | 2025-11-24 |
| 7 | Gemini 3.1 Pro | 22.55 | unverified | 2026-02-19 |
| 8 | Gemini 3 Pro | 18.6 | unverified | 2025-11-18 |
| 9 | Claude Sonnet 4.5 | 14.71 | unverified | 2025-09-29 |
| 10 | GPT-5.1 | 13.73 | unverified | 2025-11-13 |
| 11 | Gemini 3 Flash | 9.8 | unverified | 2025-12-17 |
| 12 | o3 | 8.8 | unverified | 2024-12-20 |
| 13 | Claude Opus 4 | 6.9 | unverified | 2025-05-22 |
| 14 | GPT-5 | 6.9 | unverified | 2025-08-07 |
| 15 | Kimi K2 (Jul 2025) | 4.9 | unverified | 2025-07-11 |
| 16 | Claude Sonnet 4 | 4.9 | unverified | 2025-05-22 |
| 17 | Claude 3.5 Sonnet (October 2024) | 4.6 | unverified | 2024-10-22 |
| 18 | Gemini 2.5 Pro (Jun 2025) | 3.92 | unverified | 2025-06-05 |
| 19 | Claude 3.7 Sonnet | 3.8 | unverified | 2025-02-24 |
| 20 | o4-mini | 3.6 | unverified | 2025-04-16 |
| 21 | o3-mini | 1.3 | unverified | 2025-01-31 |
| 22 | GPT-4o (Nov 2024) | 0 | unverified | 2024-05-13 |