SeamlessM4T v2 Large
Single 2.3B model for speech and text translation across 96 languages.
About
SeamlessM4T v2 Large is Meta's unified translation model, released in November 2023. A single 2.3-billion-parameter model handles speech-to-speech, speech-to-text, text-to-speech and text-to-text translation plus automatic speech recognition, covering 101 languages for speech input, 96 for text input and output, and 35 for speech output.
It pairs a w2v-BERT 2.0 speech encoder with an NLLB-style text decoder and a unit-based HiFi-GAN vocoder, and is supported by Hugging Face Transformers. The weights are released under CC-BY-NC 4.0, so commercial use is not permitted.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 123; Trending score 74/100 from researcher assessment.