Developer ToolsUnconfirmedUpdated 19 Oct 2026Not yet reviewed

vllm-sr/d3: 27B Multimodal Decision Model

vllm-sr/d3 is a 27B multimodal decision model that answers multiple-choice, yes/no, and scoring questions in a single call, supporting text, images, and videos, with documented performance and deployment instructions. 249 Hugging Face likes in 9 days since vllm-sr/d3 was created (27.3/day). 1 source across 1 channel.

Breakout48Low confidence ↓

Also in: Foundation Models · AI Infrastructure · Artificial Intelligence

Reading guideFact cited ·Analysis our interpretation ·Estimate forward-looking · [n] links to the source

AI contextual layer6 claims drafted by openai/gpt-oss-20b were kept because each quote was found verbatim in its source; 0 unverifiable claims were discarded. Statements marked AI await human review.

What happened

Developer relevanceAnalysis		11.3/15

Adoption velocityMeasured		2.4/10

Community momentumMeasuredInsufficient evidence		0/10

Source credibilityMeasured		10/10

Cross-source confirmationMeasuredInsufficient evidence		0/10

Evidence strengthMeasured		5/10

Time to impactEstimateInsufficient evidence		0/5

Hype penalty		0

Breakout Score		48

Breakout Score factors
Factor	Share of maximum	Points
NoveltyAnalysisRubric level 2/4 (Moderate). Heuristic (automated): first published 9 days ago.		7.5/15
Technical significanceAnalysisRubric level 3/4 (High). AI-assisted (openai/gpt-oss-20b), pending human review: Provides a single-call multimodal decision engine with documented latency and throughput on modern GPUs Evidence: “Speed: text requests take a median of 54.4 ms, requests with an image a median of 290 ms, and requests with a 10-second video a median of 500 ms, on one AMD Instinct MI325X, one request at a time with AutoModel.”		11.3/15
Developer relevanceAnalysisRubric level 3/4 (High). AI-assisted (openai/gpt-oss-20b), pending human review: Directly usable via vLLM serve, supports common media formats, and offers high throughput Evidence: “Run with vLLM vLLM serves d3 natively: vllm serve loads this repository as it is, with no conversion and no extra flags, and answers on vLLM's System One endpoint, /v1/systemone.”		11.3/15
Adoption velocityMeasuredIndependent integrations: unknown (0 pts). Hugging Face likes: 27.3 per day since creation; full marks at 1000/day, log-scaled (2.4/5).		2.4/10
Community momentumMeasuredInsufficient evidenceFewer than three periods of activity data, or no baseline activity to compare against.		0/10
Source credibilityMeasuredPrimary source present; mean source-tier weight 1.00 (primary 1.0, secondary 0.7, community 0.35).		10/10
Cross-source confirmationMeasuredInsufficient evidenceNo independent publisher has confirmed this yet.		0/10
Evidence strengthMeasuredMet 2/4: primary source ✓, runnable artifact (code, weights or package) ✓, reproducible or independent benchmark ?, technical documentation or paper ✗. 1 unknown (scored 0).		5/10
Time to impactEstimateInsufficient evidenceNo time-to-impact estimate yet.		0/5
Hype penaltyNo hype indicators detected.		0
Breakout Score		48

Factor

Points

NoveltyAnalysisRubric level 2/4 (Moderate). Heuristic (automated): first published 9 days ago.

7.5/15

Technical significanceAnalysisRubric level 3/4 (High). AI-assisted (openai/gpt-oss-20b), pending human review: Provides a single-call multimodal decision engine with documented latency and throughput on modern GPUs Evidence: “Speed: text requests take a median of 54.4 ms, requests with an image a median of 290 ms, and requests with a 10-second video a median of 500 ms, on one AMD Instinct MI325X, one request at a time with AutoModel.”

11.3/15

Developer relevanceAnalysisRubric level 3/4 (High). AI-assisted (openai/gpt-oss-20b), pending human review: Directly usable via vLLM serve, supports common media formats, and offers high throughput Evidence: “Run with vLLM vLLM serves d3 natively: vllm serve loads this repository as it is, with no conversion and no extra flags, and answers on vLLM's System One endpoint, /v1/systemone.”

11.3/15

Adoption velocityMeasuredIndependent integrations: unknown (0 pts). Hugging Face likes: 27.3 per day since creation; full marks at 1000/day, log-scaled (2.4/5).

2.4/10

Community momentumMeasuredInsufficient evidenceFewer than three periods of activity data, or no baseline activity to compare against.

0/10

Source credibilityMeasuredPrimary source present; mean source-tier weight 1.00 (primary 1.0, secondary 0.7, community 0.35).

10/10

Cross-source confirmationMeasuredInsufficient evidenceNo independent publisher has confirmed this yet.

0/10

Evidence strengthMeasuredMet 2/4: primary source ✓, runnable artifact (code, weights or package) ✓, reproducible or independent benchmark ?, technical documentation or paper ✗. 1 unknown (scored 0).

5/10

Time to impactEstimateInsufficient evidenceNo time-to-impact estimate yet.

0/5

Hype penaltyNo hype indicators detected.

Breakout Score

vllm-sr/d3: 27B Multimodal Decision Model

What happened

vllm-sr/d3: 27B Multimodal Decision Model

What happened

Why it matters

Should I care?

What changed?

Who is affected?

What can developers do with it?

Is it production-ready?

What would make it important?

What could prevent adoption?

Technical change

Architecture

Developer impact

Adoption & momentum

Timeline

Limitations & risks

Competing approaches

Who is building on it

Evidence & sources

Primary source (1)

Secondary source (0)

Community signal (0)

How the score of 48 was calculated

What happened

What happened

Why it matters

Should I care?

What changed?

Who is affected?

What can developers do with it?

Is it production-ready?

What would make it important?

What could prevent adoption?

Technical change

Architecture

Developer impact

Adoption & momentum

Timeline

Limitations & risks

Competing approaches

Who is building on it

Evidence & sources

How the score of 48 was calculated

Related signals