Sophon-Realtime v1: Tag and Flash

Sophon-Realtime v1: open full-duplex voice models that hear your tone and speak with one.

Three builds share one brain; they differ in how the reply becomes sound.

	Tag (recommended)	Flash	E2E (preview)
Voice	expressive speech model performs sound tags	lighter, faster speech model	talker inside the 32B model
NVV-SuperBench en / zh	0.80 / 0.86	0.63 / 0.77	0.60 / 0.77
First audio, one user	~0.6 s	0.31 s	~1 s
Users per 80 GB GPU	16	40	4-5
GPUs	1	1	2

• Tag: HF / ModelScope / code in tag/
• Flash: HF / ModelScope / code in flash/
• E2E: research preview (two GPUs): HF / ModelScope / guide in docs/deploy.md

Also: pip install sophon-realtime (Tag build + sophon-realtime serve), and Live mode (OpenAI GPT-Live compatible gateway) in live/.

Apache-2.0 code and weights.