Horangi 4 - GLP Average: leaderboard
General language performance (GLP) average of the Horangi Korean LLM Leaderboard 4 (W&B Korea), over 11 categories covering syntax, semantics, expression, retrieval, knowledge, reasoning, function calling and coding. 0-100, higher is better; its Expression category averages Ko-MT-Bench's stored score, which is a 1-10 rating divided by 100 on most runs.
Metric: Score (%). Source: horangi.ai. Saturation forecast: Around February 2027. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 (High) | 83.93 | #20 (Claude Fable 5) |
| 2 | Gemini 3.1 Pro (Preview) (High) | 83.43 | #64 (Gemini 3.1 Pro (Preview)) |
| 3 | GPT-5.5 (xHigh) | 82.76 | #33 (GPT-5.5) |
| 4 | GPT-5.6 Sol (Max) | 82.73 | #22 (GPT-5.6 Sol) |
| 5 | Gemini 3.5 Flash | 82.27 | #66 |
| 6 | GPT-5.4 (2026-03-05) (xHigh) | 81.87 | #63 (GPT-5.4 (2026-03-05)) |
| 7 | GPT-5.6 Terra (Max) | 81.58 | #40 (GPT-5.6 Terra) |
| 8 | Gemini 3 Pro (Preview) (High) | 80.48 | #71 (Gemini 3 Pro (Preview)) |
| 9 | Claude Opus 4.6 (High) | 80.29 | #67 (Claude Opus 4.6) |
| 10 | Grok 4.5 | 80.17 | #54 |
| 11 | Claude Sonnet 5 (High) | 79.31 | #65 (Claude Sonnet 5) |
| 12 | Qwen 3.7 Max | 78.82 | #85 |
| 13 | GPT-5.6 Luna (Max) | 78.51 | #79 (GPT-5.6 Luna) |
| 14 | Qwen 3.7 Plus | 78.15 | #98 |
| 15 | GPT-5.2 (2025-12-11) (xHigh) | 77.77 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=horangi-4-glp-average · How It Works · Data refreshed daily, snapshot 2026-11-05.