AI Battle > Rankings > AA-WER (Non-streaming): AA-AgentTalk Dataset
null by null leads the AA-WER (Non-streaming): AA-AgentTalk Dataset ranking with a score of 0.02. The ranking covers 20 evaluated speech-to-text, re-scored on every AI Battle telemetry run.
Last updated: . Source: AI Battle live evaluation telemetry, 20 models ranked.
| # | Model | Creator | Score |
|---|---|---|---|
| 1 | Scribe v2, ElevenLabs | 0.02 | |
| 2 | Gemini 3.1 Pro Preview (High) | 0.02 | |
| 3 | Qwen3.5 Omni Plus | 0.02 | |
| 4 | Gemini 3.1 Pro Preview (Low) | 0.02 | |
| 5 | MAI-Transcribe-2 | 0.02 | |
| 6 | Smallest AI Pulse Pro | 0.02 | |
| 7 | Universal-3.5 Pro, AssemblyAI | 0.02 | |
| 8 | MAI-Transcribe-1.5 | 0.02 | |
| 9 | Fun-Realtime-ASR-preview | 0.02 | |
| 10 | Gemini 3.5 Transcribe | 0.02 |
null by null currently leads the AA-WER (Non-streaming): AA-AgentTalk Dataset ranking with a score of 0.02, based on live AI Battle telemetry across Speech-to-Text.
The AA-WER (Non-streaming): AA-AgentTalk Dataset ranking is refreshed automatically from the AI Battle evaluation cluster. Scores are re-evaluated on every telemetry run, so positions reflect the latest published results.
The top 3 are: 1. null (0.02), 2. null (0.02), 3. null (0.02).