GigaChat 2 Max: benchmark results
Sber's GigaChat 2 Max, the flagship model in their GigaChat 2.0 line for the Russian language, with a 128,000-token context window and document analysis up to 200 pages. Provider: Sber. Released 2025-03-13. Access: API.
Aggregate Rating 1570 ± 28, rank #545 of 1635 rated models, from 43 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA - ruWorldTree | 99.62 | Accuracy (%) | 98.6 |
| MERA - ruTiE | 90.07 | Accuracy (%) | 92.4 |
| MERA - USE | 58.14 | Grade, normalized (%) | 90.5 |
| MERA - ruOpenBookQA | 94.75 | Accuracy (%) | 89.3 |
| MERA - PARus | 93 | Accuracy (%) | 87.7 |
| MERA - ruModAr | 96.32 | EM (%) | 87.2 |
| MERA - MaMuRAMu | 86.37 | Accuracy (%) | 86.2 |
| MERA - ruHumanEval | 50.24 | pass@1 (%) | 86.1 |
| MERA - ruCodeEval | 50.55 | pass@1 (%) | 85.7 |
| MERA - CheGeKa | 41.39 | F1 (%) | 84.4 |
| MERA - ruMMLU | 79.51 | Accuracy (%) | 83.8 |
| MERA - BPS | 99.3 | Accuracy (%) | 82.1 |
Interactive version: theaggregate.ai/model?slug=gigachat-2-max · How It Works · Data refreshed daily, snapshot 2026-11-05.