Can model X hold seat Y?

Every recorded change to the outbound fleet's seat policy from 23 July to 15 September, placed on the throttling record that was being logged at that moment. Step through the changes and the pools move exactly as the readings say they did.

Eight frozen cells across three beds replayed Fable 5.1 at @high and @xhigh into the DECLARER, ITERATE, and FIND review seats — blind, on the exact diff the incumbent saw — against terra, gpt-5.5, and the GLM swarm. Fable is DECLARER/FIND-shaped: deep single-seam verify-after-fix, honest verdicts, zero fabricated cites in eight cells, and bugs terra missed. It is not ITERATE-shaped. @xhigh beat @high on every bed.

We asked whether Spark could occupy a judgment seat in a frozen-cell benchmark. This run produced a narrower answer: the candidate exhausted its weekly capacity before the bench could say anything about capability.

Run alongside the Spark bench on the same frozen cells, an 8-seat glm-5.3 review swarm was scored against the incumbent 4-seat sonnet swarm at round-1 review. It matches or beats the incumbent on opus-critical coverage on every cell, at higher precision and zero fabricated citations — while the sonnet swarm fails the bench's own citation circuit-breaker.

gpt-5.6-terra leads a full build from plan lock to convergence, scored blind against the shipped Claude Opus build of the same unit. Checklist conformance was a dead heat; the separation came from gate truthfulness, stalls, and scope completion.

GLM Coding Pro (6× quota) landed 2026-08-21. The owner ordered n≥2 testing of every candidate seat with a hard-fail override; 21 GLM trials + 2 opus comparators settled which new seats GLM can hold.

Nine scored cells against shipped ground truth. glm-5.3 reads real code with zero hallucinated citations, but in every cell it misses or under-weighs one must — so the seats it can take are the ones a stronger confirm still backstops.

DeepSeek burns no subscription quota, so it is the only pool that can absorb work when both subscriptions are hot. Benched on the GLM sandboxes, it earns four seats where an automatic oracle checks it — and is barred from every review seat, where it finds musts but mis-weights them.