652 views · 8 likes · 1 reposts (at collection)
d3-ultra debuts as an open 27B multimodal decision model
- Who
- vLLM Semantic Router team
- What
- released d3-ultra, an open 27B multimodal decision model
What happened
vLLM highlights d3-ultra, which @XunzhuoLiu introduces as a new open 27B multimodal decision model ranked #2 of 230 on the Jev Decision Index text board, +3.03 over d3. vLLM describes per-answer probabilities, visual grounding, and video tracking, with training and serving on AMD Instinct GPUs. Apache-2.0 weights: https://huggingface.co/vllm-sr/d3-ultra
Read next
See the full storyvLLM Semantic Router Team Releases d3-ultra, an Open 27B Multimodal Decision ModelThe story so far
The vLLM Semantic Router team released d3-ultra, a 27B multimodal decision model, with open weights available under the Apache-2.0 license. vLLM said the model ranks second among 230 models on the Jev Decision Index text board, scoring 3.03 points above d3. vLLM described d3-ultra as providing per-answer probabilities, visual grounding and video tracking, and said training and serving use AMD Instinct GPUs.
Keep up with AI, one email a day
Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue · 2026-10-27