MESH - Allocentric Dynamic Transcription: leaderboard
MESH tests spatial hearing across synchronized receivers and moving speakers. Direction, distance, position and transcription tasks remain separate. Word error rate can exceed 100%. Overall gives equal weight to localization and per-question transcription quality; it is not an average of the printed task metrics. This board reports corpus WER (%). Lower is better.
Metric: Corpus WER (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2027. 10 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | MESH-agent | Evaluation run: V2; Qwen3-14B planner and source specialist tools | 59.15 | |
| 2 | Qwen 3.5 Omni Plus (Harness: MESH FOA-sep) | 68.91 | ||
| 3 | Qwen3.8 Omni Flash | 73.68 | #115 | |
| 4 | Qwen 3.5 Omni Plus (Harness: MESH FOA-W) | 74.1 | ||
| 5 | MiDasheng-Spatial | 81.18 | ||
| 6 | Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-sep) | 92.06 | ||
| 7 | Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-W) | 95.1 | ||
| 8 | BAT | 99.98 | ||
| 9 | JAEGER Classical | 101.38 | ||
| 10 | JAEGER Neural | 101.57 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mesh-allocentric-dynamic-transcription · How It Works · Data refreshed daily, snapshot 2026-11-16.