JP-TL-Bench: leaderboard
Metric: LT Score (0-10; both directions, anchor-set Bradley-Terry fit, gemini-2.5-flash judge). Source: arxiv.org. Saturation forecast: Around October 2026. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 9.94 |
| 2 | Gemini 2.5 Flash | 9.89 |
| 3 | Qwen 3 30B A3B 2507 Instruct | 9.63 |
| 4 | GPT-4o | 9.12 |
| 5 | Llama 3.3 70B Instruct | 6.72 |
| 6 | Phi-4 | 5.12 |
| 7 | Qwen 3 4B | 4.1 |
| 8 | LFM2 2.6B | 3.91 |
| 9 | Llama 3.1 8B Instruct | 2.94 |
| 10 | Phi-4 Mini Instruct | 0.99 |
| 11 | Llama 3.2 3B Instruct | 0.54 |
| 12 | LFM2-350M | 0.14 |
| 13 | TinySwallow-1.5B | 0.04 |
Interactive version: theaggregate.ai/benchmark?slug=jp-tl-bench · How It Works · Data refreshed daily, snapshot 2026-09-25.