All pull requests

Pull requests

• Add GPT-6.1 Terra (release, 51 models) [WIP]

  #247·MaxGhenis opened Oct 27, 2026Contributor·

  ·
  2

• Tool pilot: a web+pe3 arm, web search with the version 3 calculator

  #246·MaxGhenis opened Oct 27, 2026Contributor·

  ·
  1

• Confirmatory runner: lanes, one engine pool per model, admission and the ceiling

  #244·MaxGhenis opened Oct 25, 2026Contributor·22 of 25 tasks·

  ·
  1

• Run v2 on fresh Microcosm households in judged batches

  #240·MaxGhenis opened Oct 22, 2026Contributor·

  ·
  1

• Leave recorded ceiling-exhaustion rows alone when applying an adjudication

  #224·MaxGhenis opened Oct 13, 2026Contributor·

  ·
  1

• Keep a judged case's parse-failure label to cases where a model returned no value (d1514)

  #223·MaxGhenis opened Oct 13, 2026Contributor·

  ·
  1

• Seal each release's tag at the commit that holds its board

  #211·MaxGhenis opened Oct 10, 2026Contributor·

  ·
  1

• Call the October upgrade's fixed outputs fixed, and build the table's split from a helper

  #209·MaxGhenis opened Oct 10, 2026Contributor·

  ·
  1

• Pin the reference adversary's other scripts to the pass's inputs

  #207·MaxGhenis opened Oct 9, 2026Contributor·

  ·
  2

• Pin the reference adversary's leaderboard impact to the pass's inputs

  #205·MaxGhenis opened Oct 9, 2026Contributor·

  ·
  1

• Install PolicyBench from GitHub in the README quick start

  #199·MaxGhenis opened Oct 6, 2026Contributor·

  ·
  1

• Declare the Unlicense everywhere, matching LICENSE

  #198·MaxGhenis opened Oct 6, 2026Contributor·

  ·
  1

• Add a reusable unlisted-input sweep and keep local taxes and credits out of the state outputs

  #196·MaxGhenis opened Oct 5, 2026Contributor·

  ·
  3

• Harden the GPT-6.1 Sol driver: pin kept sidecars, reference explanations, re-opened flags and the export receipt

  #190·MaxGhenis opened Oct 4, 2026Contributor·

  ·
  3

• Probe card-less models under the card the onboarding gauntlet derives

  #189·MaxGhenis opened Oct 3, 2026Contributor·

  ·
  1

• Record PolicyEngine provenance without importing policyengine

  #186·MaxGhenis opened Sep 30, 2026Contributor·

  ·
  1

• Compute PolicyEngine provenance once per chunked or retry invocation

  #184·MaxGhenis opened Sep 29, 2026Contributor·

  ·
  2

• Count only real request timeouts in the supervisor's concurrency backoff

  #183·MaxGhenis opened Sep 29, 2026Contributor·

  ·
  1

• Merge batch repair rounds cell by cell, like sync

  #180·MaxGhenis opened Sep 28, 2026Contributor·

  ·
  1

• Add an unactivated v2 household prompt contract for issue 165

  #173·MaxGhenis opened Sep 7, 2026Contributor·

  ·

• Validate publication cost provenance with a frozen offline contract

  #172·MaxGhenis opened Sep 7, 2026Contributor·

  ·
  1

• Add a launchd launcher for supervised runs and document what kills them

  #163·MaxGhenis opened Sep 5, 2026Contributor·

  ·
  1

• Stability measurement suite: answer, reasoning, and counterfactual layers (spec + rung-0 implementation)

  #159·MaxGhenis opened Aug 25, 2026Contributor·

  ·
  1

• Temporarily hide leaderboard costs

  #119·MaxGhenis opened Jul 9, 2026Contributor·

  ·
  1

• Refresh PolicyBench references and rankings

  #54·MaxGhenis opened May 23, 2026Contributor·

  ·
  1