All collections

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。

Open in model catalog Compare models

24 models in this collection. Preview models show catalog availability; open a model to check its gateway and provider details.

24 models

Google: Gemini 3.5 Transcribegoogle/gemini-3.5-transcribeContext98.3KInputSpecialized pricingOutputSee model pricing

Fish Audio: Transcribe 1 Profish-audio/transcribe-1-proContextN/AInputSpecialized pricingOutputSee model pricing

AssemblyAI: Universal-3.5 Proassemblyai/universal-3-5-proContextN/AInputSpecialized pricingOutputSee model pricing

Meta: Muse Voice Transcribe 1.0meta/muse-voice-transcribe-1.0ContextN/AInputSpecialized pricingOutputSee model pricing

Microsoft AI: MAI-Transcribe 2microsoft/mai-transcribe-2ContextN/AInputSpecialized pricingOutputSee model pricing

NVIDIA: Nemotron 3.5 ASR Streaming Multilingual 0.6Bnvidia/nemotron-3.5-asr-streaming-multilingual-0.6bContextN/AInputSpecialized pricingOutputSee model pricing

Mistral: Voxtral Small 24B 2507 STTmistralai/voxtral-small-24b-2507-sttContextN/AInputSpecialized pricingOutputSee model pricing

Mistral: Voxtral Mini 3B 2507mistralai/voxtral-mini-3b-2507ContextN/AInputSpecialized pricingOutputSee model pricing

Qwen: Qwen3 ASR 1.7Bqwen/qwen3-asr-1.7bContextN/AInputSpecialized pricingOutputSee model pricing

Qwen: Qwen3 ASR 0.6Bqwen/qwen3-asr-0.6bContextN/AInputSpecialized pricingOutputSee model pricing

OpenAI: GPT Transcribeopenai/gpt-transcribeContextN/AInputSpecialized pricingOutputSee model pricing

Fish Audio: Transcribe 1fish-audio/transcribe-1ContextN/AInputSpecialized pricingOutputSee model pricing

How models enter this collection

Membership uses declared transcription output and does not claim word-error-rate rankings.

Membership comes from public reference metadata. CLSSAI callability is checked against its public directory.