Add gpt-5.5-pro competitor + partial-coverage asterisk in leaderboard#27
Merged
Conversation
Add openai/gpt-5.5-pro ($30/$180) as an OpenRouter raw-api-loop competitor. A cost-capped probe (4 of 9 cases, $91.30) found it has a spiky edge: it uniquely cracked GHSA-9f49 (CWE-416 use-after-free) that 0/16 other competitors found, but missed a case a cheaper model got — not a default-include at ~9-20x the cost-per-case. Because pro ran a partial denominator, its detection rate (2/4 = 50%) is not rank-comparable with full-corpus competitors. Flag any competitor that completed fewer than the fullest run's case count with a footnoted asterisk in the HTML leaderboard (on both the name and the Detect cell), so a partial-coverage rate is not misread as a rank. Principled, no hardcoded names: the two qwen 8/9 runs are flagged too. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Pull request overview
Adds a new OpenRouter competitor entry for openai/gpt-5.5-pro and updates the HTML leaderboard to visually flag competitors whose results are based on fewer audited cases than the fullest run, to prevent partial-coverage detection rates being misread as rank-comparable.
Changes:
- Add
raw-api-loop/gpt-5.5-protocompetitors.example.yamlwith OpenRouter pricing metadata. - Compute the maximum audited case count across leaderboard entries and append a footnoted asterisk to the Competitor name and Detect cells for partial-coverage rows.
- Add an explanatory footnote to the leaderboard when partial-coverage entries are present.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| nelson/html_report.py | Adds partial-coverage detection-rate asterisk markers and a footnote to the leaderboard HTML output. |
| competitors.example.yaml | Adds raw-api-loop/gpt-5.5-pro competitor configuration (model id, auth profile, pricing). |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
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
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.
Add openai/gpt-5.5-pro ($30/$180) as an OpenRouter raw-api-loop competitor. A cost-capped probe (4 of 9 cases, $91.30) found it has a spiky edge: it uniquely cracked GHSA-9f49 (CWE-416 use-after-free) that 0/16 other competitors found, but missed a case a cheaper model got — not a default-include at ~9-20x the cost-per-case.
Because pro ran a partial denominator, its detection rate (2/4 = 50%) is not rank-comparable with full-corpus competitors. Flag any competitor that completed fewer than the fullest run's case count with a footnoted asterisk in the HTML leaderboard (on both the name and the Detect cell), so a partial-coverage rate is not misread as a rank. Principled, no hardcoded names: the two qwen 8/9 runs are flagged too.