vLLM news
The latest vLLM news, : Baidu releases Unlimited-OCR-2 with unlimited-horizon parsing via SR-SWA.
8 stories mentioning vLLM since 16 September 2026, newest first, each with its sources and discussion. Follow to see new ones on your front page.
-
Baidu releases Unlimited-OCR-2 with unlimited-horizon parsing via SR-SWA
Baidu released Unlimited-OCR-2 on October 27, 2026, introducing Sparse Reference Sliding Window Attention (SR-SWA) that compresses text-visual attention interactions to a fixed-size set of working references such as…
1 source primary sourcehuggingface.co -
IBM and Red Hat launch llm-d 1.0 distributed inference stack
IBM Research announced llm-d 1.0 at the PyTorch Conference North America on October 20, 2026. The Kubernetes‑native stack is positioned as a vendor‑neutral way for organizations to run any model on any accelerator in…
3 sourcesMarkTechPostUnite.AItechtarget.com -
vLLM 0.31 adds constrained decoding for 100% tool-call accuracy
Ubicloud found that open-weight models like Kimi K2.6 and DeepSeek V4 Flash often fail to generate correct tool calls, with vLLM version 0.28 achieving only 92% schema accuracy for Kimi K2.6. This led to nonsensical…
1 source HN 47ubicloud.com -
Arm says its AGI CPU delivers nearly 2× performance per rack of AMD EPYC 9755
Arm announced that its AGI CPU, built for agentic AI workloads, shows a modeled 1.99x rack‑level performance advantage over AMD’s EPYC 9755 when evaluated across seven workloads. The claim comes from Arm‑commissioned…
1 sourceTom's Hardware -
Intel releases LLM-Scaler-vLLM PV 1.5
Intel software engineers released LLM-Scaler-vLLM PV 1.5, a Docker stack that runs large‑language models on Intel Arc Graphics. It also adds support for several Qwen and Gemma models, including Qwen3.6‑27B,…
1 sourcephoronix.com -
openKylin unveils KART runtime and agent environment at OSS EU 2026
openKylin, an open‑source Linux community from China, showcased its new KART unified AI runtime and a system‑level agent execution environment during the Open Source Summit Europe 2026 in Prague. Executive Director Min…
1 sourceMarkTechPost -
Google Antigravity SDK adds support for local Gemma 4 26B models
Google has announced that the Antigravity SDK now supports local AI workflows, enabling developers to run agentic capabilities entirely offline. The initial release features support for the Gemma 4 26B A4B model using…
1 sourcedevelopers.googleblog.com -
Model Serves Memory: KV Cache Tax in AI Inference
A mid-sized model deployment fell over due to a memory issue when under real traffic, despite having enough weights and compute resources. The key-value (KV) cache was the culprit, growing with each concurrent request.…
1 sourceTowards Data Science
Questions about vLLM
What is the latest vLLM news?
As of 27 October 2026: Baidu releases Unlimited-OCR-2 with unlimited-horizon parsing via SR-SWA. Baidu released Unlimited-OCR-2 on October 27, 2026, introducing Sparse Reference Sliding Window Attention (SR-SWA) that compresses text-visual attention interactions to a fixed-size set of working references such as 512 tokens.