vLLM Semantic Router releases d3-ultra, a 27B open multimodal decision model
d3-ultra is a 26.09B-parameter open multimodal decision model from vLLM Semantic Router. It takes text or JSON with optional images and videos, reads the input once per request, and answers multiple questions (choice, yes/no, score, box, track) with probabilities without generating text. Context is 131,072 tokens per question, licensed Apache-2.0. The page reports latency of 304 ms to 2.16 s for 1 to 255 questions on one AMD Instinct MI325X.
Open original source huggingface.coOne primary source; other links are context, not independent corroboration.
- Primary sourceX discovery · automated brief
- Discovery linkX
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
- Original repository+40
- Feed text or abstract+15
- No editor link review0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
d3-ultra is a 26.09B-parameter open multimodal decision model from vLLM Semantic Router. It takes text or JSON with optional images and videos, reads the input once per request, and answers multiple questions (choice, yes/no, score, box, track) with probabilities without generating text. Context is 131,072 tokens per question, licensed Apache-2.0. The page reports latency of 304 ms to 2.16 s for 1 to 255 questions on one AMD Instinct MI325X.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: huggingface.co. Brief basis: Hugging Face model card. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Source claims not independently checked; latency figures are the publisher's own benchmark on a single hardware configuration.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- The model reads the input once and answers every question on it, returning a probability for each option.
- Supports choice, yes/no, score, box (images and video moments) and track (whole videos) decision types.
- Context window is 131,072 tokens per question; licensed Apache-2.0.
- Reported median latency is 304 ms to 2.16 s for 1 to 255 questions, measured on one AMD Instinct MI325X.
Source claims not independently checked; latency figures are the publisher's own benchmark on a single hardware configuration.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
These are Grok Bot discovery notes. AINewsIQ prepared a brief from available source material. The bot's rationale, linked posts, authors, and stances have not been independently verified.
The vLLM Semantic Router team published d3-ultra (Apache-2.0, 26.09B params incl. a 0.46B vision encoder, 131,072-token context per question) on Hugging Face. It reads text, JSON, images or video once and returns probabilities for choice, yes/no, score, box and track questions instead of generating text, served on one AMD Instinct MI325X. A free API is offered. Latency figures are self-reported.
Source claims not independently checked; latency figures are the publisher's own benchmark on a single hardware configuration.
Engagement counts, when shown, are observed snapshots. They do not establish accuracy or corroboration.
- huggingface post ↗ · 20 likes, 56 downloads · observed 10/27/2026, 4:01:58 AM
- x post ↗ · 11 likes, 1 reposts, 1 replies, 0 quotes · observed 10/27/2026, 4:01:48 AM
- x post ↗ · 4 likes, 0 reposts, 3 replies, 2 quotes · observed 10/27/2026, 4:01:21 AM
- Related post ↗
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
Grok Bot submitted the link. AINewsIQ fetched the primary source and prepared a source-based brief automatically. An editor has not reviewed the story; social attention and the brief do not independently verify its claims.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from Hugging Face model card.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
An open-weights model that answers many structured questions about one multimodal input in a single pass could cut cost and latency for classification and routing pipelines compared with per-question generation.
Limits and open questions
Source claims not independently checked; latency figures are the publisher's own benchmark on a single hardware configuration.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.