Fetching from the wire…
01
02
03
04
05
06
07
08
Infra2026-07-19 · source-backed
As of July 13, AMD Quark trains, quantizes, and serves EAGLE-3 drafts through vLLM, reporting up to 2.00x for Kimi-K2.5 and 1.79x for MiniMax-M2.5. Related July work: ROCm end-to-end on MI355X with a prebuilt container (July 10), plus Hopper-optimized attention and FP8 MoE backends improving TTFT and TPOT on H20. If you self-host and haven't turned on a draft model, this is the biggest single-config win currently sitting on the table. (vLLM Blog)
Each link below shares sources, entities, or timing with this story.
Kimi built by Moonshot AI / Shared entity: Kimi / Earlier coverage
Linked by a graph relationship (Kimi built by Moonshot AI); both cover Kimi; earlier Kimi coverage from 2026-03-21.
Kimi built by Moonshot AI / Shared entity: MiniMax / Earlier coverage
Linked by a graph relationship (Kimi built by Moonshot AI); both cover MiniMax; earlier MiniMax coverage from 2026-03-17.
Linked by a graph relationship (Kimi built by Moonshot AI); both cover MiniMax; earlier MiniMax coverage from 2026-03-04.
Linked by a graph relationship (Kimi built by Moonshot AI); both cover MiniMax; earlier MiniMax coverage from 2026-03-02.
Linked by a graph relationship (Kimi built by Moonshot AI); both cover MiniMax; earlier MiniMax coverage from 2026-02-24.
MiniMax competes with Anthropic / Shared entity: July / Earlier coverage
Linked by a graph relationship (MiniMax competes with Anthropic); both cover July; earlier July coverage from 2026-07-16.
Linked by a graph relationship (MiniMax competes with Anthropic); both cover July; earlier July coverage from 2026-07-16.
Linked by a graph relationship (MiniMax competes with Anthropic); both cover July; earlier July coverage from 2026-07-12.