AssemblyAI STT-3.7: three speech-to-text models for live calls, voice agents and recordings
A three-model speech-to-text family — pre-recorded, realtime, and a voice-agent-tuned realtime variant — covering 32 languages with native code-switching, from $0.21/hr.
Why it matters
One family split by job: the voice-agent variant posts 4.53% WER on 12,460 real calls, the lowest of the 26 streaming models AssemblyAI benchmarked.
What you could build with it
A clinic could run a voice agent that books appointments in 32 languages, transcribing each call live and handing the caller to a human when the request gets complicated.
Does it hold up?
Usable now and already the default for new accounts; the 4.53% voice-agent WER is vendor-measured, so teams should test it on their own call audio before switching.
Built with AssemblyAI STT-3.7
- AssemblyAI vs Deepgram (2026): accuracy, pricing, limitsassemblyai · Head-to-head on the 3.7 family vs Deepgram across recorded, live and voice-agent workloads.
- The voice agent launch runbook: ramp, rollback, and on-callassemblyai · Production runbook updated for STT-3.7-Realtime-Agent, with canary and rollback guidance.
- Multilingual voice agents: 32 languages, native code-switchingassemblyai · Report on the 3.7 family's 32-language coverage and LiveKit/Pipecat reference implementations.
Learn more
First spotted on article: source.
More AI models
Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Qwen4-Max: Alibaba's new flagship model goes liveAlibaba's 2.4T-parameter flagship is now generally available via API, with a 1M-token context, native vision, and…models · JEV 0.67smolmaxxing: run 16B LLMs on a 1 GB TV box or a 2013 iPadAn MIT-licensed llama.cpp patch set that runs language models up to a 16B MoE on devices with about 1 GB of usable RAM…models · JEV 0.65Blue TTS v3: a 388MB open text-to-speech model with voice cloningAn open-source 388MB text-to-speech model (MIT) with voice cloning from about 10 seconds of speech, 7 languages, 15…models · JEV 0.65Haru-4BOpen full-duplex spoken-dialogue foundation model that listens and speaks simultaneously (7.5 Hz frames, 133 ms per…models · JEV 0.63
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.