SeamlessM4T v2
Massively multilingual speech-to-speech translation — 101 languages, speech in, speech out.
About
SeamlessM4T v2 (November 2023) is Meta's foundational speech translation model: speech-to-speech and speech-to-text translation across 101 languages, plus text-to-speech and ASR in a unified architecture. It preserves vocal style and prosody across translation better than cascaded pipelines.
Licensing is the catch: CC-BY-NC 4.0 — research/non-commercial only. Commercial products cannot use these weights. For research, low-resource language work, and prototyping multilingual voice experiences, it remains the reference open model. Note the weights are large (v2-Large) and inference needs a solid GPU. Meta has not released a commercially-licensed successor.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-claimed (Meta paper): state-of-the-art on Fleurs S2ST and ASR benchmarks at release — treat as provider-claimed; paper reports standardized benchmark numbers (Fleurs, CVSS). Independent: widely cited as the open S2ST reference.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier). Technically the reference open speech-to-speech translation model with paper-reported benchmark numbers, but capped hard: (1) CC-BY-NC 4.0 bans commercial use outright — the single biggest limitation; (2) 2023-era, no updates or commercial successor from Meta; (3) heavy inference requirements. Essential for research and low-resource languages; unusable for products.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.1/10
- Maintenance · recency, releases, issue load
- 4.0/10
- Openness · license + docs
- 5.2/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · HF downloads 303,272 (peer-normalized)
- · released 1048 days ago
- · manual review: reference open S2ST model with paper benchmarks; 2023-era, no commercial successor; license: CC-BY-NC 4.0 — research/non-commercial only