other
0.039B parameters
Commercial OK
Reviewed May 2026

Whisper Tiny

Smallest member of the Whisper encoder-decoder ASR family (39M params). Trained on 680k hours of weakly supervised multilingual audio. Targets sub-realtime transcription on CPU and tiny edge devices; ships in transformers, whisper.cpp, faster-whisper and ONNX/CoreML/TensorRT variants.

License: apache-2.0·Context: 30 tokens
BLK · VERDICT

Our verdict

OP · Fredoline Eruo|VERIFIED MAY 29, 2026
unrated

The default 'good enough' baseline when latency and footprint matter more than WER. If accuracy matters, jump to distil-large-v3 or Parakeet.

Overview

Smallest member of the Whisper encoder-decoder ASR family (39M params). Trained on 680k hours of weakly supervised multilingual audio. Targets sub-realtime transcription on CPU and tiny edge devices; ships in transformers, whisper.cpp, faster-whisper and ONNX/CoreML/TensorRT variants.

Strengths

  • Runs realtime on Raspberry Pi 4 / mobile CPUs via whisper.cpp
  • 99-language coverage out of the box
  • Apache-2.0 weights, no commercial restrictions
  • Mature ecosystem (faster-whisper, WhisperX, ONNX exports)

Weaknesses

  • WER ~15-20% on LibriSpeech test-other vs ~3% for large-v3
  • Hallucinates words on silence and music
  • 30-second context window forces chunking with overlap for long audio
  • Weak on heavy accents, code-switching, and overlapping speakers

Quantization variants

Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.

QuantizationFile sizeVRAM required
Q4_K_M0.0 GB1 GB

Get the model

HuggingFace

Original weights

huggingface.co/openai/whisper-tiny

Source repository — direct quantization required.

Hardware that runs this

Cards with enough VRAM for at least one quantization of Whisper Tiny.

Compare alternatives

Models worth comparing

Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.

Step down
Smaller — faster, runs on weaker hardware
No verdicted models in the next tier down yet.

Frequently asked

What's the minimum VRAM to run Whisper Tiny?

1GB of VRAM is enough to run Whisper Tiny at the Q4_K_M quantization (file size 0.0 GB). Higher-quality quantizations need more.

Can I use Whisper Tiny commercially?

Yes — Whisper Tiny ships under the apache-2.0, which permits commercial use. Always read the license text before deployment.

What's the context length of Whisper Tiny?

Whisper Tiny supports a context window of 30 tokens (about 0K).

Source: huggingface.co/openai/whisper-tiny

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.

Related — keep moving

Before you buy

Verify Whisper Tiny runs on your specific hardware before committing money.