Qwen 3.8 Flash Next rumored: 125B + 51B N-gram lookup tables, launches tomorrow
teortaxesTex · x · 2026-08-25
Rumors say Qwen 3.8 Flash Next (125B params + 51B N-gram embeddings, 6B active, based on next-gen Qwen 4 architecture) launches tomorrow. superelesha analyzes the 51B N-gram tables: a massive trainable lookup-memory that hashes recent token N-grams and mixes fetched vectors into the hidden state, so the main model spends less effort on frequent local patterns.
Memory math (ideal 4bpw):
- 125B main weights ≈ 58.2 GiB
- 51B N-gram tables ≈ 23.7 GiB, 82 GiB total
- Leaves only 14 GiB spare on 4×3090s
Caveat: real Q4 is usually heavier, and some quantizers keep embedding tables in FP16/BF16 — the dtype of the 51B parameters is the key question of this release.
More from Infra
- NVIDIA launches Jetson Orin Nano 2: 2x performance, 40% less power — nvidia · 2026-08-26
- Studies Reveal Data Centers Boost Local Jobs and Wages Significantly — justin_hart · 2026-08-26
- Opinion: Land Scarcity Drives Shift to Decentralized AI Compute — bittingthembits · 2026-08-26
- Jalapeno beats VR200 with optimized DeepSeek implementation, faster execution — itsclivetime · 2026-08-26
- CUDA code now runs on Apple Silicon with zero source code changes — petewoodbridge · 2026-08-26
- Sail Research CEO on inference efficiency for long-running AI agents, from chips to engines — agihouse_org · 2026-08-26