MESH - Allocentric Dynamic Transcript Position: 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 native localization success (%). Higher is better.

Metric: Native localization success (%). Source: arxiv.org. Saturation forecast: Around 2030. 10 models tracked. Updated 14 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1MESH-agentEvaluation run: V2; Qwen3-14B planner and source specialist tools18.39
2Qwen 3.5 Omni Plus (Harness: MESH FOA-W)6.93
3Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-W)6.48
4Qwen3 Omni 30B A3B Instruct (Harness: MESH FOA-sep)6.45
5Qwen 3.5 Omni Plus (Harness: MESH FOA-sep)6.41
6Qwen3.8 Omni Flash4.19#115
7MiDasheng-Spatial0.19
8JAEGER Neural0
9JAEGER Classical0
10BAT0

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-transcript-position · How It Works · Data refreshed daily, snapshot 2026-11-16.