Google launches Gemini 3.5 Transcribe with 2.6% WER and 0.4s latency
ArtificialAnlys · x · 2026-08-27
Google released Gemini 3.5 Transcribe APIs, including a pre-recorded version and a Live streaming version. Key metrics:
- Non-streaming: Achieves 2.6% AA-WER (Rank #5) and processes at 84x realtime.
- Streaming: First final transcript arrives 0.40s after speech end (4.0% WER); first partial arrives in 0.25s.
- Features: Supports 85+ languages, custom vocabulary, auto-formatting, and multi-speaker ID (up to 3).
- Pricing: $5/1k minutes for Transcribe, $9/1k minutes for Live.
Related event: Google Launches Gemini 3.5 Transcribe Speech-to-Text Model(29 posts)→
More from Multimodal
- Higgsfield Genjutsu hits 4B monthly reach, opens API at 50% off — SimplyAnnisa · 2026-10-13
- Genjutsu videos top 4B views, yet no mobile app exists — SimplyAnnisa · 2026-10-13
- Reddit user hunts for a model that can nail this anime art style, asks if LoRA training is the answer — Amjad_mj · 2026-10-13
- Generating a 30-second AI music video with Seedance 2.5 Max on Pixio — tsi_org · 2026-10-13
- AI video characters now actually perform: expressions, timing and dialogue improve in months — jamestagg · 2026-10-13
- Open ComfyUI workflow paints full UV textures for MakeHuman characters from one image — o0ANARKY0o · 2026-10-13