Continuity: a tie and a gap on the same bank

2026-09-14 · Council of AI (CSOAI Ltd)

Three MEASURED cards on one continuity bank: two tie exactly, and a third sits well above them without a separation test.

On the continuity axis, three signed pod cards bind the same bank_sha256, 25dd6a72ac5a, each at n=33. ollama:qwen3:8b reads 0.6667 at n=33 (https://councilof.ai/interop/mill-cards-signed/signed-continui-70d556baa835.json). ollama:gemma3:4b reads 0.303 at n=33 (https://councilof.ai/interop/mill-cards-signed/signed-continui-948290b6734e.json). ollama:qwen2.5:1.5b reads 0.303 at n=33 (https://councilof.ai/interop/mill-cards-signed/signed-continui-287505bc7ea9.json). The last two tie exactly, at 10 correct of 33 each; the qwen3:8b card records 22 correct of 33. All three bodies say MEASURED with zero parse errors excluded. So this one bank carries a tie and a gap at the same time, and both are reported as they are. The gap between the qwen3:8b card and the two tied cards has no separation test attached; it is a difference between point estimates. The tie is a tie at the resolution of 33 items, not a claim that the two tied models behave identically elsewhere.

How to verify: fetch the artifacts listed above yourself, and check a signed card against the published key at https://councilof.ai/gspc-verify/.

Measurement, not certification. Not a grade, endorsement or legal finding.