Audio

MITOpen weights where releasedUpdated August 2026

Whisper Large V3 Turbo

Whisper Large V3 Turbo is a Whisper-family option for speech recognition, transcription, subtitles, meeting notes, and multilingual audio workflows.

OpenAI · Whisper

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026SourcesExact model card

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Fast transcription and multilingual audio workflows

Who should use it

  • Fast transcription and multilingual audio workflows
  • Builders who want local or self-hosted testing options.

Common workflows

  • Fast transcription and multilingual audio workflows
  • audio workflows
  • transcription workflows
  • speech recognition workflows
  • local workflows

Deployment and hardware notes

~1.6 GB in fp16 — half of large-v3 — so an 8 GB card handles it with room for batching, and int8 brings it under a gigabyte.

License and usage notes

MIT. Open weights where released. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

  • Open weights where released model option for Whisper workflows.
  • Fast transcription and multilingual audio workflows
  • The usual default for interactive or real-time-ish transcription: pick it first, and only step up to large-v3 if measured accuracy on your own audio falls short.

Limitations

  • The trimmed decoder gives up some accuracy on hard audio (heavy accents, overlapping speakers, noisy recordings), and OpenAI notes it was not trained for speech translation the way large-v3 was, so translation quality is the weak spot.
  • ~1.6 GB in fp16 — half of large-v3 — so an 8 GB card handles it with room for batching, and int8 brings it under a gigabyte.
  • Context window and limits: 30-second audio windows · 448-token cap per window.
  • Verify the exact model card, provider docs, license, and serving support before production use.

Local workflow notes

The usual default for interactive or real-time-ish transcription: pick it first, and only step up to large-v3 if measured accuracy on your own audio falls short.

Local runtimes: Transformers, faster-whisper, whisper.cpp

Platforms: Windows, macOS, Linux

Transcription spec

Memory~1.6 GB in fp16 (809M parameters) · ~0.8 GB in int8Decoder depth4 layersLanguagesMultilingual (99 languages)Audio window30-second audio windows · 448-token cap per window

Large-v3 with the decoder cut from 32 layers to 4 and then fine-tuned — the encoder is untouched, which is why quality stays close to large-v3 on most audio while decoding is several times faster.

Sources to verify

Exact model card

Related resources

Continue with model source notes, local tools, and implementation guides related to this model.

Whisper family hubAI tools directoryModel source notesModels directoryGuides library
Hardware~1.6 GB in fp16 (809M parameters)RuntimeTransformers, faster-whisper, whisper.cppContext30-second audio windows · 448-token cap per windowLast updated2026
Exact model card →

Model ecosystem connections

Use these next-step links to move from this profile into related tools, comparisons, guides, stacks, and curated shortlists.

Recommended runtimes and tools

Transformers →

Setup and deployment

Check if this model fits your GPU →What can 12GB VRAM run? →What is quantization? GGUF, Q4, Q8 explained →What is VRAM and how much do you need? →

Related model pages

Whisper family hub →Whisper Medium →Whisper Small →Whisper Tiny →Distil-Whisper Large V3 →Faster-Whisper Large V3 →Whisper Base →

Guides, stacks, and comparisons

All model comparisons →

Related models

Whisper MediumBuilders balancing local transcription quality and runtime cost for meetings, media, and batch audio processing.Whisper SmallLocal transcription setups that need a lighter model for captions, notes, and general speech-to-text tasks.Whisper TinyFast prototypes, edge-style experiments, and low-memory ASR tests where accuracy tradeoffs are acceptable.Distil-Whisper Large V3Builders who want strong multilingual transcription quality with a lighter checkpoint for production-style speech pipelines.Faster-Whisper Large V3Teams optimizing Whisper-style batch or local transcription pipelines where runtime efficiency and deployment control matter.Whisper BaseDevelopers who want a practical starting point for local captioning, voice notes, and CPU-leaning ASR experiments.

Ready to run this model locally?

Find a compatible interface in our Local AI Tools directory →