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

#ModelSettingScoreOverall rank
1MESH-agentEvaluation run: V2; Qwen3-14B planner and source specialist tools59.15
2Qwen 3.5 Omni Plus (Harness: MESH FOA-sep)68.91
3Qwen3.8 Omni Flash73.68#115
4Qwen 3.5 Omni Plus (Harness: MESH FOA-W)74.1
5MiDasheng-Spatial81.18
6Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-sep)92.06
7Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-W)95.1
8BAT99.98
9JAEGER Classical101.38
10JAEGER Neural101.57

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.