Boost Inference Performance up to 15x on NVIDIA Blackwell Using DFlash Speculative Decoding - NVIDIA Developer
NVIDIA Developer publishes the DFlash speculative-decoding technique, claiming up to 15x inference throughput gains on Blackwell GPUs.
AI Storyline
3 items · 3 sources · 3 days
Operational story trace
Speculative Decoding
Follow in this browser to see new updates on your Live feed.
Current stateSpreading across hardware vendors
Latest change
vLLM and AMD Quark shipped EAGLE-3 speculative decoding for AMD Instinct GPUs on Jul 13, reporting up to 2.00x throughput for Kimi-K2.5 and 1.79x for MiniMax-M2.5 — the first non-NVIDIA vendor to validate speculative decoding as a production latency lever.
Earlier contextThe story so far
Speculative decoding — drafting candidate tokens ahead of the target model to cut serving latency — is a known lever for LLM inference cost. NVIDIA's June 23 developer post introduced DFlash, a speculator built on the target model's own KV projections, claiming up to 15x throughput gains on Blackwell GPUs, and Modal adopted it in production days later with its own open-source speculator models.
editor-curated · source-linked
State over time
● launch · Jun 23second vendor (AMD) · Jul 13 → now ●
- launch · Jun 23
- adopted (NVIDIA) · Jun 24 → Jul 12
- second vendor (AMD) · Jul 13 → now
LAUNCH · Jun 23NVIDIA introduces DFlash, claiming up to 15x inference gains on Blackwell1 source · scout · show source ▾
ADOPTED · Jun 24Modal ships open-source DFlash speculator models with Z Lab and SGLang1 source · scout · show source ▾
Achieve state-of-the-art inference latencies with speculative decoding
Modal names DFlash as the technique behind its own state-of-the-art latency results, releasing open-source speculator models built with Z Lab and SGLang.
NOW · Jul 13AMD Instinct gets its own speculative-decoding path via EAGLE-3 and vLLM1 source · show source ▾
EAGLE-3 Speculative Decoding on AMD Instinct GPUs: Training and Serving with vLLM and AMD Quark
AMD and vLLM ship EAGLE-3 speculative decoding tuned with AMD Quark for Instinct GPUs, the first non-NVIDIA vendor validation of the pattern.
Day 1 Tuesday, Jun 23, 2026
Boost Inference Performance up to 15x on NVIDIA Blackwell Using DFlash Speculative Decoding - NVIDIA Developer
NVIDIA Developer publishes the DFlash speculative-decoding technique, claiming up to 15x inference throughput gains on Blackwell GPUs.
Day 2 Wednesday, Jun 24, 2026
Achieve state-of-the-art inference latencies with speculative decoding
Modal names DFlash as the technique behind its own state-of-the-art latency results, releasing open-source speculator models built with Z Lab and SGLang.
Day 3 Monday, Jul 13, 2026
EAGLE-3 Speculative Decoding on AMD Instinct GPUs: Training and Serving with vLLM and AMD Quark
AMD and vLLM ship EAGLE-3 speculative decoding tuned with AMD Quark for Instinct GPUs, the first non-NVIDIA vendor validation of the pattern.
What to watch — open questions
- Does DFlash's speedup hold on non-Blackwell GPUs, or is it Blackwell-specific?
- Will DFlash itself get ported to AMD/vLLM, or does EAGLE-3 stay the AMD-side technique?
How this thread was built
scout surfaced 2editor wrote the arc · 3 beatswatcher 1 status change
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.