Sber: kandinsky
1 benchmarks
Leaderboard position
#925
Overall rank (of 1121 models)
149
Elo rating
1
Benchmarks
Top 82.5%
Percentile
In Sber (of 13 models)
Above: Black Forest Labs: flux.2-klein-base-9b #924
Below: NVIDIA Corporation: parakeet-ctc-0.6b #926
Benchmark highlights
Best benchmark
llmarena.ru - Text-to-Image (Snapshot from july 6, 2026)
905
Worst benchmark
llmarena.ru - Text-to-Image (Snapshot from july 6, 2026)
905
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 1 of 1 benchmarks
1/1
100.0%
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| MATH-500 | ~74.15 | 50% | — |
| AIME 2025 | ~42.96 | 50% | — |
| AIME 2026 | ~59.85 | 50% | — |
| HMMT Feb 2026 | ~51.31 | 50% | — |
| GPQA Diamond | ~42.56 | 50% | — |
| MMLU-Pro | ~43.91 | 50% | — |
| Humanity's Last Exam (no tools) | ~4 | 50% | — |
| IFEval | ~72.64 | 50% | — |
| IFBench | ~41.66 | 50% | — |
| HumanEval | ~53.23 | 50% | — |
| MBPP | ~50.21 | 50% | — |
| Terminal-Bench v2.1 | ~8.6 | 50% | — |
| MMLU | ~49.49 | 50% | — |
| SWE-bench Verified | ~17.06 | 50% | — |
| SWE-bench Pro | ~5.24 | 50% | — |
| SciCode | ~15.9 | 50% | — |
| AA-LCR | ~29.31 | 50% | — |
| Artificial Analysis Intelligence Index v5.0.1 | ~12 | 50% | — |
| Humanity's Last Exam (with tools) | ~11.9 | 50% | — |
| GSM8K | ~56.29 | 50% | — |
| -(CoderPPL) | ~-5.29 | 50% | — |
| LiveCodeBench v6 | ~39.82 | 50% | — |
| MMMU-Pro | ~39.71 | 50% | — |
| MATH | ~31.15 | 50% | — |
| BBH | ~49.67 | 50% | — |