ESF-Bench - Multi-Source: leaderboard
Metric: Tagged slot accuracy (%; slots tagged with a MS (Multi-Source) scenario, multi-hop inference across persona, knowledge base, session and form sources, including conflicts; 810 synthetic multi-turn enterprise slot-filling samples (6,530 slots, 8 domains) whose prompts combine a schema, a user-assistant conversation and auxiliary sources; varied output formats mapped to a standard form before exact comparison with the ground truth; 32k maximum tokens, reasoning setting as printed per model). Source: arxiv.org. Saturation forecast: Around December 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 32B (Thinking) | 87.3 |
| 2 | Gemini 2.5 Flash | 86.9 |
| 3 | GPT-5.1 (High) | 84.1 |
| 4 | GPT-OSS-120B (High) | 80.8 |
| 5 | Ministral-3-14B-Reasoning-2512 | 78.4 |
| 6 | GPT-OSS-120B (Low) | 75.9 |
| 7 | GPT-5.1 (Non-reasoning) | 73.1 |
| 8 | Gemini 2.5 Flash (Non-reasoning) | 71.9 |
| 9 | Ministral-3-14B-Instruct-2512 | 66.9 |
| 10 | Qwen 3 32B (Non-reasoning) | 66.1 |
Interactive version: theaggregate.ai/benchmark?slug=esf-bench-multi-source · How It Works · Data refreshed daily, snapshot 2026-09-29.