Skip to content
AI News HubLIVE
More
Summary only

vLLM Semantic Router Releases d3-ultra: A 27B Open Multimodal Decision Model

vLLM Semantic Router has released d3-ultra, a 27B Apache-2.0 multimodal decision model that does not generate text but instead reads an input once and returns probabilities for every answer to every question, scoring 87.2 on the team's own SR-Text test.

SourceMarkTechPostAuthor: Asif Razzaq
vLLM Semantic Router Releases d3-ultra: A 27B Open Multimodal Decision Model
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • d3-ultra has 26.09B parameters, a Qwen3.5-style architecture, 131,072-token context and is released under Apache-2.0.
  • It never generates text; it reads the input once and returns probabilities for choice, yes/no, score, box and track questions.
  • It leads the team's private SR-Text test at 87.2 versus GLiDE V1.2 at 85.1, but trails d3-turbo on SR-Box grounding, 58.2% versus 60.0%.
  • All benchmark results are self-reported on private test sets, and the model was tested on one AMD Instinct MI325X with BF16 weights.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Models