APEX-Agents
Rank #1 of 51 · 39 models scored · top score 49.6
unknown
Strongest on discrimination, weakest on saturation headroom. 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 | Gemini 3.5 Flash | 49.6 | unverified | 2026-05-19 |
| 2 | Claude Fable 5 | 45 | unverified | 2026-06-09 |
| 3 | Claude Opus 4.8 | 42.5 | unverified | 2026-05-28 |
| 4 | GPT-5.5 | 38.4 | unverified | 2026-04-23 |
| 5 | GPT-5.4 | 35.9 | unverified | 2026-03-05 |
| 6 | GLM-5.2 | 35.6 | unverified | 2026-06-16 |
| 7 | GPT-5.2 | 34.3 | unverified | 2025-12-11 |
| 8 | Claude Opus 4.7 | 33.9 | unverified | 2026-04-16 |
| 9 | Gemini 3.1 Pro | 33.5 | unverified | 2026-02-19 |
| 10 | Claude Opus 4.6 | 31.7 | unverified | 2026-02-05 |
| 11 | GPT-5.3 Codex | 31.7 | unverified | 2026-02-05 |
| 12 | GPT-5.4 Mini | 24.6 | unverified | 2026-03-17 |
| 13 | Gemini 3 Flash | 24 | unverified | 2025-12-17 |
| 14 | Claude Sonnet 4.6 | 23.7 | unverified | 2026-02-17 |
| 15 | Kimi K2.6 | 18.9 | unverified | 2026-04-20 |
| 16 | Gemini 3 Pro | 18.4 | unverified | 2025-11-18 |
| 17 | Claude Opus 4.5 | 18.4 | unverified | 2025-11-24 |
| 18 | GPT-5 | 18.3 | unverified | 2025-08-07 |
| 19 | GPT-5.1 | 17.5 | unverified | 2025-11-13 |
| 20 | o3 | 17.2 | unverified | 2024-12-20 |
| 21 | GLM-5 | 17.2 | unverified | 2026-02-11 |
| 22 | GPT-5.4 Nano | 16.9 | unverified | 2026-03-17 |
| 23 | Grok 4 | 15.2 | unverified | 2025-07-09 |
| 24 | Kimi K2.5 | 14.4 | unverified | 2026-02-02 |
| 25 | Qwen 3.5 Plus (hosted 397B-A17B) | 13.6 | unverified | 2026-02-16 |
| 26 | Gemini 3.1 Flash-Lite | 12.8 | unverified | 2026-03-03 |
| 27 | Claude Sonnet 4 | 9.3 | unverified | 2025-05-22 |
| 28 | Claude Haiku 4.5 | 8.9 | unverified | 2025-10-15 |
| 29 | DeepSeek-V3.2 | 7 | unverified | 2025-12-01 |
| 30 | Gemini 2.5 Pro (Jun 2025) | 6.6 | unverified | 2025-06-05 |
| 31 | MiniMax-M2.5 | 6.2 | unverified | 2026-02-12 |
| 32 | gpt-oss-120b | 4.7 | unverified | 2025-08-05 |
| 33 | Kimi K2 Thinking | 4 | unverified | 2025-11-06 |
| 34 | GLM-4.7 | 3.1 | unverified | 2025-12-22 |
| 35 | GLM-4.6 | 3 | unverified | 2025-09-30 |
| 36 | Grok 3 | 2.1 | unverified | 2025-02-17 |
| 37 | Gemini 2.5 Flash (Jun 2025) | 1.8 | unverified | 2025-06-17 |
| 38 | o1 | 1.1 | unverified | 2024-12-05 |
| 39 | GPT-4o (Nov 2024) | 1.1 | unverified | 2024-05-13 |