Pyyan / Compare / Hy4 preview vs Qwen3.8-Flash-Next
Hy4 preview vs Qwen3.8-Flash-Next
2 of 5Open-Weight Models · verified 2 Sept 2026
Hy4 previewTencentcurrentQwen3.8-Flash-NextAlibabacurrent
| Specification | Hy4 preview | Qwen3.8-Flash-Next |
|---|---|---|
| Summary | 770B parameters under Apache 2.0, and it helped optimise its own training. | The most downloaded thing on Hugging Face right now, and a stated Qwen4 preview. |
| Context | 1M | 262K native, ~1M extended |
| Input ($/Mtok) | $0.834 | $0.15 |
| Output ($/Mtok) | $2.501 | $0.47 |
| Licence | Apache 2.0 | Open weights |
| Max output | 64K | 32K |
| Parameters | 770B total, 49B active | 125B total, 6B active |
| Released | 28 Aug 2026 | 26 Aug 2026 |
| Model ID | tencent/Hy4-preview | Qwen/Qwen3.8-Flash-Next |
| Category | Open-Weight Models | Open-Weight Models |
| Official | — | — |
Highlighted rows are where these differ.
Hy4 preview
- 770B total parameters with 49B active, 78 layers, 256 routed experts plus one shared
- Apache 2.0, with an FP8 quantised build shipped alongside it
- Tencent used it during its own development, and report a 31.8% inference throughput gain from it
Best for the largest open weights.
Qwen3.8-Flash-Next
- 125B total parameters with only 6B active per token
- Alibaba describes it as a preview of the Qwen4 architecture
- Top of the Hugging Face trending list, with a GGUF build close behind it
Best for open weights at speed.