vLLM Semantic Router and AMD release Vela 3.0: 380M model beats GPT-5-mini at zero-shot
vllm_project · x · 2026-10-28
The vLLM Semantic Router, KR Labs and AMD teams launched Vela 3.0, a collection of six open routing models (0.3B-27B) that return routing decisions and named span extractions in a single call.
Highlights:
- vela3-fast: a 0.4B UltraBERT encoder holding the highest public Decision Index up to 500M
- Beats GPT-5-mini at zero-shot classification
- 18 ms per span question; the vision variant runs decisions and spans on images with OCR checks at 0.25B per call, CPU-friendly
- Six variants (fast/vision/lite/mini/flash/turbo) available as a Hugging Face collection with browser demo
More from Infra
- Row-Bot: a local 27B model runs a shop's morning ops entirely on one PC via MCP — Acceptable-Object390 · 2026-10-28
- Streaming inference stack runs full-precision DeepSeek V4.1 on one 5090 at 1770 tok/s prefill — Front_Eagle739 · 2026-10-28
- Richard Socher's Recursive open-sources Flint: 40B MoE, 2.5B active, Apache 2, runs coding locally on Mac — altryne · 2026-10-28
- Cantor: SK hynix trades at just 3x CY27 EPS as AI cycle proves durable — firstadopter · 2026-10-28
- Wall Street gripped by new anxiety as it bankrolls AI's vast data center buildout — shiringhaffary · 2026-10-28
- Netflix runs observability on RDF/OWL/SHACL ontology, 38M log events per second — JeremyCMorgan · 2026-10-28