Qwen3.8 vs 3.6 Benchmark: 3.8 is 5-20% slower but holds quality
Radiant_Condition861 · reddit · 2026-08-15
A benchmark on RTX PRO 6000 Blackwell comparing Qwen3.8-27B-FP8 and Qwen3.6-27B-FP8 using MTP Sweep shows Qwen3.8 is 5-20% slower across all steps. However, quality remains comparable or slightly better in some cases. If raw speed is the priority, 3.6 remains superior, especially at higher MTP steps.
More from Infra
- Three CoreWeave co-founders have sold $3.3B in shares since IPO amid credit worries — SumitGup · 2026-10-16
- Cherry pi on a Raspberry Pi: the tiny-model rumor developers are watching — sull · 2026-10-16
- CoreWeave's debt costs falling: DDTL spreads tightened from S+9.62% to S+2.25% — BenBajarin · 2026-10-16
- SoftBank lines up US IPOs for SB Energy and AI robotics spinoff Roze, valuations top $50B — nordicinst · 2026-10-16
- TRL reproduces Modal's mixed-precision async RL: plain GRPO collapses, SC+MIS holds 0.72 — QGallouedec · 2026-10-16
- Hugging Face scales Indic pretraining with 2.8T tokens and an open data pipeline — Hugging Face Blog · 2026-10-16