tencent/Hy-SimulMT-30B-A3B-Interleave
just landedmodel
Hy-SimulMT is a real-time simultaneous translation model developed by Tencent Hunyuan, supporting bidirectional speech-to-speech translation between Chinese and English. Mixture-of-experts, 30B total, 3B active (128 experts, 8 per token). Context: 256K tokens (config). License: apache-2.0. Vendor-reported: "As this small-scale capability assessment shows, Hy-SimulMT achieves translation quality on par with expert human simultaneous interpreters, while often responding faster (with lower latency)."
- Vendor-reported: As this small-scale capability assessment shows, Hy-SimulMT achieves translation quality on par with expert human simultaneous interpreters, while often responding faster (with lower latency).
- Org
- tencent
- Released
- 2026-10-22
- Status
- just landed
- License
- apache-2.0 (permissive)
- Parameters
- 30.4B total, 3.0B active
- Architecture
- MoE · 128 experts, 8/token · HYV3SpeechForCausalLM
- Context
- 262,144 tokens
- Task
- translation · text
- Signal
- 4 HF likes · 0 downloads
Runs on
real quant files; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~21.5 GB: 24GB GPU (3090 / 4090), 32GB GPU (5090), 24GB GPU (7900 XTX), Mac 36GB, CPU only, 32GB RAM (slow)
- Q8 ~35.1 GB: 48GB GPU (RTX 6000 Ada / A6000), Mac 48GB, Mac 64GB (M1 Max), CPU only, 64GB RAM (slow)
- FP16 ~62.6 GB: 96GB GPU (RTX Pro 6000 Blackwell), 1× DGX Spark (128GB), 2× DGX Spark (256GB), Strix Halo 128GB unified (~96GB usable), Mac 96GB, Mac 128GB, Mac 192GB, Mac 256GB, Mac 512GB, CPU only, 128GB RAM (slow), CPU only, 256GB RAM (slow)
Links
First published here 2026-10-22 · updated 2026-10-26 · markdown · JSON