Pending benchmarksTracking since Oct 18, 2026
Is Agens Volundr 32B v1 state of the art?Pending benchmarks.Announced Oct 18, 2026; no independent results yet.
Blockway's open-weights 32B model for agentic coding, the full release after its Preview, tuned for long agent sessions.
Announced facts
The details so far
- Lab
- Blockway
- Announced
- Oct 18, 2026
- API
- Not yet
- Open weights
- Published
- Parameters
- About 32B
- Focus
- Agentic coding, tool use and reasoning
- Weights
- Hugging Face
- License
- Apache-2.0
- Local serving
- Agens Engine (MIT), Blockway's single-GPU inference engine
- API
- Earlier said to be coming; not confirmedunsure
Tags marked unsure are not stated on the lab's own pages.
Next steps
What would let us call it
Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.
- Publicly announced with a source
- Usable by the public (API or weights)
- Listed on an independent board we track
- Enough results for a rank in a category
- Holds across two snapshots (crowns only)
Self-reported
What the lab claims
Self-reported. Shown for context; not used in any verdict.
Lab claim (self-reported): repetition loops on its 50-task SWE-bench set went from 36/50 (Preview) to 0/50.
Lab claim: runs a 262K-token context within a 16 GB GPU memory budget on Agens Engine.
Receipts
Sources
LABBlockway announcement on Xx.com/blockwaylink/status/2111743226034208871 · Oct 18, 2026REPOAgens-Volundr-32B-v1 on Hugging Facehuggingface.co/Blockway/Agens-Volundr-32B-v1 · Oct 18, 2026Tracking continues until an independent board lists it.