Status after #277: one of three items done, and the repro no longer
reproduces — which is not the same as fixed
Re-ran the issue's exact repro against beast on the deployed build
(74df8fd):
…
max_completion_tokens, so the output cap is unenforced for current OpenAI clients
limit.output is a KV planning reserve advertised as an output ceiling — nothing reconciles the two
limit.output is a KV planning reserve advertised as an output ceiling — nothing reconciles the two
Landed and verified — closing, with one item handed to #223
Merged 74df8fd0, deployed, checked against beast and through cortex.
The advertisement is now the ceiling:
neuron …
max_completion_tokens, so the output cap is unenforced for current OpenAI clients
Fixed and verified — closing
Landed as part of #278 (74df8fd0) rather than the branch described
above: the four generation loops now take the cap from
effective_max_tokens() (current…
Verified live — closing
Merged 6945cf98, deployed to the fleet, probed against beast's
Qwen3.8-27B on 74df8fd.
The test plants a fact that exists only inside a replayed reasoning item…
limit.output is a KV planning reserve advertised as an output ceiling — nothing reconciles the two
A real client already reads this field, and our value would break it
Checked against the DSH/pi-ai source. pi-ai's provider discovery parses
an OpenAI-compatible GET /models listing and…
A case for what "saturation" has to mean here
An investigation on 2026-08-24 needed exactly the signals this issue proposes to export, could not get them from a dashboard, and had to reconstruc…
A real run of this benchmark, by hand, with numbers
The operator ran the tower-defence prompt across three models through
DSH on 2026-08-24 (~/git/tower-defense: opus5/,
`deepseek-v4-flash/…