Skip to content

Russian

Русский
Joint highest accuracy
gpt-oss-120bOpenAI · High reasoning
96.0%

3 results share the top score

A close alternative
Qwen3.7 FlashQwen · High reasoning
96.0%

Matches the highest score

How much scores vary
Lowest score72.0%Highest score96.0%
24.0pp

Between the highest and lowest of 11 published results.

How the models compare

Published results, ranked by accuracy. Higher is better.

Compare side by side
Russian · How the models compare
#ModelAccuracyGap to top
1gpt-oss-120bHigh reasoning
gpt-oss-120b: 96.0%96.0%
Top score
1Qwen3.7 FlashHigh reasoning
Qwen3.7 Flash: 96.0%96.0%
Top score
1Qwen3.7 FlashLow reasoning
Qwen3.7 Flash: 96.0%96.0%
Top score
2Ling 3.0 FlashHigh reasoning
Ling 3.0 Flash: 92.0%92.0%
-4.0 pp
2Ling 3.0 FlashLow reasoning
Ling 3.0 Flash: 92.0%92.0%
-4.0 pp
2Solar Pro 4High reasoning
Solar Pro 4: 92.0%92.0%
-4.0 pp
2Solar Pro 4Low reasoning
Solar Pro 4: 92.0%92.0%
-4.0 pp
3gpt-oss-120bLow reasoning
gpt-oss-120b: 88.0%88.0%
-8.0 pp
3Ling 3.0 FlashMedium reasoning
Ling 3.0 Flash: 88.0%88.0%
-8.0 pp
3Mercury 2.5 PreviewHigh reasoning
Mercury 2.5 Preview: 88.0%88.0%
-8.0 pp
4Mercury 2.5 PreviewLow reasoning
Mercury 2.5 Preview: 72.0%72.0%
-24.0 pp

The numbers, with the working attached.

Tests, prompts, settings and model responses live on the release pages.

View published releases