Granite Speech 3.3 8B
IBM's open Apache-2.0 speech-language model: two-pass ASR plus translation, topping the Hugging Face Open ASR leaderboard.
About
Granite Speech 3.3 8B is IBM's open speech-language model, built by modality-aligning Granite 3.3 8B Instruct to speech. It uses a two-pass design: the first call transcribes audio to text, and a second explicit call runs the underlying language model over the transcript. Everything is Apache-2.0, and IBM has topped the Hugging Face Open ASR leaderboard with it.
It handles ASR in English, French, German, Spanish, and Portuguese, plus speech translation from English into those languages as well as Japanese and Mandarin. Community-compiled leaderboard data puts it near 5.85% WER on English, competitive with the best open ASR models. IBM also ships a 2B variant and a WebGPU/ONNX build, and the model is servable through watsonx.ai, Ollama, and LM Studio.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- ~5.85% WER on the Hugging Face Open ASR Leaderboard, near the top for English (community-compiled, 2026)
- Superior performance over Whisper-style baselines on transcription and translation (provider-reported, IBM)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Apache-2.0 license and April 2025 release of the 3.2 generation from the Hugging Face model card; Granite Speech 3.3 8B launch and leaderboard-topping claim from IBM's official announcement; ~5.85% EN WER from community-compiled 2026 leaderboard analysis. Strengths: fully open weights from a stable enterprise vendor, leaderboard-class accuracy, 2B and WebGPU variants for constrained deployment. Docked for: (1) the headline leaderboard figures are provider-announced rather than independently reproduced; (2) ASR language coverage is narrower (5 languages) than multilingual competitors. License: Apache-2.0.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · Granite Speech 3.3 8B via IBM announcement; 3.2 released 2025-04-02 per HF model card
- · ~5.85% EN WER on HF Open ASR Leaderboard (community-compiled 2026); IBM claims Whisper-style baseline wins
- · 2B variant + WebGPU/ONNX builds; served on watsonx.ai, Ollama, LM Studio, Replicate