← Back to Vibers
Company
Prompt Driven
1 Intel
Intel
8-bit Qwen3.8-27B decodes 1.7x faster than BF16, slower at 16K context(github.com/promptdriven/pdd)
A single session measurement of an 8-bit converted 27B checkpoint against its bfloat16 source on one laptop: smaller on disk, faster to decode, and worse at long prefill and time to first token.