Publish dashboard-data-20260822: 32-model board — Grok 4.6 eighth, Ox Alpha preview fourth - #156
Merged
Conversation
grok-4.6: config, gauntlet-confirmed tool card (600s timeout, thinking budget), lock entry, site meta (released 2026-08-12). ox-alpha: config + provisional card for the cloaked OpenRouter preview; sensitivity/ox-alpha-2026-08.md records the 84.2 (#4 of 31) preview run beside the board, not in it. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Full 100-household run $8.70, exact 82.8 (#7), 1.9 points past grok-4.5. Judge audit 656/656 verdicts, zero suspects, strict gate clean, incumbents byte-identical. Pointer, dataset description, hero label, manifest, and the re-pinned annotations move in this commit. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MaxGhenis
changed the title
Ox Alpha: site meta (display 'Ox Alpha (preview)', listed 2026-08-21), card and config reframed from preview-beside-the-board to a dated preview row, note updated for the 32-model board. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
… Alpha preview fourth Both folded onto the live 30 in one driver: 658/658 verdicts, zero suspects, strict gate clean, incumbents byte-identical. grok-4.6 82.8 (#8, $8.70); ox-alpha 84.2 (#4, $0 in its free window) as a labeled preview row. Pointer, 32-model description, manifest, and re-pinned annotations move in this commit. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MaxGhenis
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds two models to the live board in one publish.
Grok 4.6 (xai/grok-4.6, $2/$6 per 1M, released 2026-08-12): gauntlet 3/3 + 16/16 (heavy reasoner → 600s timeout, thinking budget); 100/100 households, $8.70; exact 82.8, #8 of 32, $0.087/household. Beats grok-4.5 (80.9) by 1.9.
Ox Alpha (
stealth/ox-alpha, cloaked OpenRouter preview listed 2026-08-21, free window): publicly callable under the identical request, so it joins as a labeled preview row — display "Ox Alpha (preview)", dated by its listing, no named provider (site renders the no-logo fallback). 100/100 households at $0; exact 84.2, #4 of 32, 1,984/1,984 parsed. Write-up with the per-program table insensitivity/ox-alpha-2026-08.md. If it later ships under its own name, that model onboards normally with this run as its early read.Gates: judge audit 658/658 verdicts, 0 hedged, 0 suspects (473 carried over, 185 judged fresh); strict gate 0 errors; incumbents byte-identical. Release dashboard-data-20260822: payload + all-32 predictions + the Ox preview file.
Pointer/hero/manifest/annotations move in one commit per convention. Paper refreeze follows in its own PR.
🤖 Generated with Claude Code