AGIEval (English subset) — Explicit English-only subset of bilingual AGIEval; distinct domain axis from unmarked variants
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | NVIDIA-Nemotron-3-Ultra-550B-A55B-Base-BF16 | NVIDIA | 78.7 | 1 | 2026-08-24 |
| 2 | Nemotron 3 Ultra 550B A55B Base | NVIDIA | 78.7 | 1 | 2026-06-05 |
| 3 | Nemotron 3 Super 120B A12B | NVIDIA | 77.9 | 1 | 2026-08-11 |
| 4 | Kimi K2 Base | Moonshot | 72.5 | 1 | 2026-06-05 |
| 5 | Qwen3.5 35B A3B | Alibaba | 70.5 | 1 | 2026-08-11 |
| 6 | DeepSeek-V3.2-Exp-Base | DeepSeek | 70.1 | 1 | 2026-06-05 |
| 7 | GLM-4.5-Base | Z.ai | 70.1 | 1 | 2026-06-05 |
| 8 | NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 | NVIDIA | 70 | 1 | 2026-08-24 |
| 9 | Nemotron 3.5 Lightning | NVIDIA | 70 | 1 | 2026-08-11 |
| 10 | Mistral Large 3 675B Base 2512 | Mistral | 69.3 | 1 | 2026-06-05 |
| 11 | NVIDIA Nemotron 3 Nano 30B A3B | NVIDIA | 68.5 | 1 | 2026-08-11 |
| 12 | Gemma 4 26B A4B | 55.3 | 1 | 2026-08-11 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.