Inco AI unveils DFlash 2 parallel speculative decoding
DFlash 2 boosts output per verification pass by over 20% with ~1% added latency, gains 16–25% across benchmarks. With the new Qwen3.8-27B drafter, SGLang serves at 2.7–3.4× autoregressive throughput at batch size 1.
1 source
Developers by email
Get an email when there's news on Developers
No news that day, no email.
More stories today
- Meta builds AI 'second brain' that learns from experts
- Snowflake shares surge 22% on AI coding momentum
- Gary Marcus critiques Musk's AI prediction shift
- Lutnick says Anthropic has patched relations with US government
- HPE lifts sales forecast on AI server demand