Qwen 3.8 Flash Next: Using Next-n-grams for efficient local inference
antirez · x · 2026-08-30
Antirez analyzed the architecture of Qwen 3.8 Flash Next, which uses a 51B table of n-grams at layer 2 to enrich token representations with a gating mechanism. This design allows masking and loading 16 small vectors from the SSD during layer 1 processing. Observers note that with the large n-grams table residing on SSD, Qwen 3.8 Flash Next with 2-bit quantization could be an optimal choice for local inference on 64GB MacBooks.
More from Infra
- Reflection AI open-sources Beam under Apache 2.0 with 40+ ecosystem partners — zhyncs42 · 2026-10-16
- NVIDIA launches RTX Spark for local AI on Windows PCs — ostrisai · 2026-10-16
- SkyRL ships Day-0 RL support for Beam with +62% LoRA decode throughput in vLLM — vllm_project · 2026-10-16
- DFlash2 delivers up to 3.03x decode-only speedup for Beam across five benchmarks — vllm_project · 2026-10-16
- vLLM's Beam serving kernels cut GB200 forward time from 8.15 to 6.47 ms at TP8 batch 128 — vllm_project · 2026-10-16
- Reflection's 501B MoE Beam hits 463 tok/s on 8x GB200 with NVFP4 at day 0 — zhyncs42 · 2026-10-16