DIRECTORY-DERIVED COLLECTION
Best Speech-to-Text and Transcription Models
用于会议、通话、字幕与语音识别的转录模型。
Current membership13Exact directory field match
CLSSAI live0Callability authority: CLSSAI directory
OpenRouter previews13Temporary metadata · external links
Models in this collection
Newest directory entries first. This order is not presented as a usage, quality, or benchmark ranking.
FFish Audio: Transcribe 1fish-audio/transcribe-1OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
XxAI: Grok STT 1.0x-ai/grok-stt-1.0OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
DDeepgram: Nova-3deepgram/nova-3OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
MMicrosoft: MAI-Transcribe 1.5microsoft/mai-transcribe-1.5OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
NNVIDIA: Parakeet TDT 0.6B v3nvidia/parakeet-tdt-0.6b-v3OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
MMistral: Voxtral Mini Transcribemistralai/voxtral-mini-transcribeOpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
QQwen: Qwen3 ASR Flashqwen/qwen3-asr-flash-2026-02-10OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
GGoogle: Chirp 3google/chirp-3OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
OOpenAI: GPT-4o Mini Transcribeopenai/gpt-4o-mini-transcribeOpenRouter preview
Context 128KOutput transcriptionInput audioPrice Specialized
OOpenAI: Whisper Large V3openai/whisper-large-v3OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
OOpenAI: Whisper Large V3 Turboopenai/whisper-large-v3-turboOpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
OOpenAI: Whisper 1openai/whisper-1OpenRouter preview
Context N/AOutput transcriptionInput audioPrice Specialized
OOpenAI: GPT-4o Transcribeopenai/gpt-4o-transcribeOpenRouter preview
Context 128KOutput transcriptionInput audioPrice Specialized
Collection contract. Source fields: architecture.output_modalities. Membership uses declared transcription output and does not claim word-error-rate rankings. 策展版本、保存、分享和个性化仍在 Workbench 跟踪。