binary-reasoning-v1 / EVAL_RESULTS-frontier.md
S3RL4's picture
Reword public difficulty table cells per naming rule
cf42db1 verified
|
Raw History Blame Contribute Delete
2.22 kB

Difficulty check against a frontier model (Claude Opus 5.5)

This file records measured frontier-model difficulty checks for RE-Lab binary-reverse-engineering tasks. It is not a leaderboard and not a calibrated pass@k band table unless explicitly labelled as one.

Method and conditions

  • Agentic screens used shell/file tools (objdump, strings, xxd, Python, Docker or direct execution where available), with a 30-minute limit.
  • Host runs are non-isolated upper bounds: tools were available, but the model was not run inside a sealed buyer sandbox.
  • Single-shot checks had no tools unless stated.
  • Times are wall-clock minutes from the measured run. Partial scores are shown when the grader records partial credit.
  • This report does not claim that a task family is "frontier-hard" unless it has a measured 2-30% band on the target model class.

Results

Task set Model / mode Result Time or score Interpretation
Public agentic flag-hunt environments, 29 Harbor environments Claude Opus 5.5 with shell/file tools 29/29 environments solved, 145/145 attempts median 44 s Practice/warm-up level for a tool-using frontier model.
Public single-shot identification sample, n=12 Claude Opus 5.5, no tools requested 7/12 correct about 58% Useful sanity check, not an RL band and not comparable to the agentic set.
Spot check of the earlier trial pack (legacy ids brv-hrd-*), 20 sampled items from 10 of 19 families Claude Opus 5.5 with shell/file tools 40/40 attempts in corrected rerun 40/40 attempts Agentic Opus 5.5 solved the sampled tasks from the earlier trial pack (40/40), so as shipped they are practice for frontier models.

What this means

  • The current public agentic pack is solvable by a tool-using frontier model; it is practice/warm-up material, not proof of a 2-30% RL band.
  • The prior "hard" label was measured and did not survive agentic testing; it is no longer sold as hard-tier material without new calibration.
  • Private task families are calibrated on fresh, never-published instances across several models (open-weight and Claude sizes); results and method are shared privately with prospective buyers.