Summary
According to a post on X by @XunzhuoLiu, AMD and the vLLM Semantic Router team have released d3-ultra, an open 27B decision model. The model handles text, images and video, draws boxes around objects it finds, tracks them over time, and gives a probability for each answer. The post says d3-ultra ranks second of 230 on the Jev Decision Index text board, 3.03 points ahead of d3, and was trained and served on AMD Instinct GPUs. Its weights are available under Apache-2.0 on Hugging Face.
AI-assisted summary based on the listed source.
What happened
d3-ultra handles text, images and video, draws boxes around what it finds and tracks objects over time, and gives a probability for each answer. It ranks second of 230 on the Jev Decision Index text board, 3.03 points ahead of d3, and was trained and served on AMD Instinct GPUs. Weights are Apache-2.0 on Hugging...
Why it matters
Developers and teams looking for open multimodal models can review the Apache-2.0 weights on Hugging Face and test them on their own tasks. For users of AMD Instinct GPUs, the post shows a model both trained and served on that hardware. Next, watch for independent benchmark results, since the ranking is so far described only in this post.
What this means for you
Hardware and robotics watchers may want to track whether this becomes a product, benchmark, or deployment signal.
Signal Intelligence
Signal Strength 92%
Technical label WATCH
Public Interest 62
Category ROBOTS & HARDWARE
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 82
Practical Impact Score 56
Novelty Interest Score 94
Consequence Score 18
Curiosity Score 16
Shareability Score 73
Why this is here
VQV surfaced this signal because it is recent, relevant to LLM Inference, connected to Scobleizer News Pulse (X).
