ArmBench-LLM v2 - Correction: leaderboard

ArmBench-LLM 5-category edition Correction score: mean accuracy across Armenian Punctuation restoration and Space Fix tasks. 0-100, higher is better.

Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Not forecast. 32 models tracked. Updated 22 Nov 2026.

Top models

#ModelScoreOverall rank
1Ayb-0.5-9B90.55
2GPT-6.1 Sol88.8#15
3GPT-5.2 Pro87.91#139
4Ayb-0.5-0.8B87
5DeepSeek V4.1 Flash86.1#69
6Gemini 2.5 Flash85.63#466
7Ayb-0.5-4B84.85
8Gemma 4 31B (IT)83.69#354
9Gemini 3 Flash (Preview)83.57#156
10GPT-5.4 Mini81.38#409
11Mistral Large 381.17#835
12Claude 3.7 Sonnet80.92#455
13Grok 4 Fast77.88#446
14Claude Sonnet 476.77#414
15Grok 4.1 Fast76.61#431

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=armbench-llm-v2-correction · How It Works · Data refreshed daily, snapshot 2026-11-22.