GLM-4-Voice
End-to-end speech language model for natural voice dialogue
About
GLM-4-Voice is an end-to-end speech language model from Zhipu AI, open-sourced on October 25, 2024. It processes speech directly as tokens without a separate STT-LLM-TTS pipeline, with a 9B model trained on a trillion speech tokens and a streaming decoder that starts generating audio from just 10 tokens. It supports Chinese and English with adjustable emotion, tone, speed and dialect. A quantized build runs on a 12GB GPU, which is unusually light for a speech LLM. Note Zhipu rebranded internationally as Z.ai in July 2025.
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 (verified tier, metrics-v1). Evidence: Zhipu AI open-sourced Oct 2024; Apache 2.0 per third-party research doc; End-to-end speech architecture with streaming decoder; Company rebranded internationally as Z.ai in July 2025. License: Apache 2.0 (per third-party research doc — confirm on repo before commercial use)..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Zhipu AI open-sourced Oct 2024; Apache 2.0 per third-party research doc
- · End-to-end speech architecture with streaming decoder
- · Company rebranded internationally as Z.ai in July 2025