AIMarkTechPostabout 3 hours ago
vLLM Releases d3-ultra Decision Model
4 min read
The vLLM Semantic Router team released d3-ultra, a 27B open multimodal decision model under Apache-2.0 on October 25, 2026. This model reads input once to answer questions, achieving an 87.2 SR-Text score. It runs on an AMD Instinct MI325X with a 304 ms latency per question. The model trails d3-turbo on Box grounding at 58.2% versus 60.0%.
Level
Hype check
Tap to vote and see what everyone thinks.
