Is vLLM-SR d3 (27B) state of the art?Yes.SOTA in Decision models (tied).
As of Oct 15, 2026, vLLM-SR d3 (27B) is tied for state of the art with low confidence in Decision models (since Oct 14, 2026; sources Decision Index Full; methodology v1.0). vLLM-SR d3 (27B) is co-SOTA in Decision models with low confidence because primary sources disagree beyond the margin of error (with linne-oak 1.0 (31B), Perplexity Decider v1.1 27B (Qwen3.8-27B, noncausal decision readout)).
The ruling, in plain English
vLLM-SR d3 (27B) is co-SOTA in Decision models with low confidence because primary sources disagree beyond the margin of error (with linne-oak 1.0 (31B), Perplexity Decider v1.1 27B (Qwen3.8-27B, noncausal decision readout)).
Reliability
Index bins for vLLM-SR d3 (27B). Confidence on the horizontal axis, accuracy on the vertical axis.
Median 97.1 ms, same GPU.
Evidence
| Source | Score | Date | Type | Receipt |
|---|---|---|---|---|
| Decision Index Full vLLM-SR d3 (27B) | 66.26 65.8–66.8 | 2026-10-14 | INDEP. | snap_decision-index_2026-10-14_release-v3|decision-index-0.3.2 |
Where it sits in the succession
Related pages
Is vLLM-SR d3 (27B) SOTA?
As of Oct 15, 2026, vLLM-SR d3 (27B) is tied for state of the art with low confidence in Decision models (since Oct 14, 2026; sources Decision Index Full; methodology v1.0).
What is vLLM-SR d3 (27B) best at?
As of Oct 15, 2026, vLLM-SR d3 (27B) is tied for state of the art with low confidence in Decision models (since Oct 14, 2026; sources Decision Index Full).
How does vLLM-SR d3 (27B) rank in Decision models?
Best current rank: #1 in Decision models, as of Oct 15, 2026.