Vals ReverseEngBench (ReverseEngBench)
Tests whether AI agents can reverse engineer real-world, contamination-free binaries.
How BenchLM shows ReverseEngBench
BenchLM mirrors the public Vals AI ReverseEngBench leaderboard captured from https://www.vals.ai/benchmarks/reverse_eng and updated by Vals on August 10, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
ReverseEngBench is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.
ReverseEngBench score on ReverseEngBench — August 10, 2026
BenchLM mirrors the published reverseengbench score view for ReverseEngBench. GPT-5.6 Sol leads the public snapshot at 30.53% , followed by Claude Opus 5 (12.21%) and GPT-5.5 (3.82%). We do not use these results to rank models overall.
GPT-5.6 Sol
OpenAI
openai/gpt-5.6-sol
Claude Opus 5
Anthropic
anthropic/claude-opus-5
GPT-5.5
OpenAI
openai/gpt-5.5
ReverseEngBench score table (5 models)
ScoreThe published ReverseEngBench snapshot places GPT-5.6 Sol first at 30.53%. The third row is 26.72 points behind. The broader top-10 range is 30.53 points, so the table still separates the published systems.
5 models have been evaluated on ReverseEngBench. The benchmark falls in the External benchmark mirrors category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. ReverseEngBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About ReverseEngBench
Year
2026
Tasks
Binary reverse-engineering tasks with full and partial-credit outcomes
Format
Accuracy and capability score
Difficulty
Real-world binary reverse engineering
BenchLM mirrors the Vals ReverseEngBench leaderboard as display-only external evidence. Rows combine models with an external agent runner and therefore do not enter weighted model rankings.
BenchLM freshness & provenance
Version
ReverseEngBench 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
FAQ
What does ReverseEngBench measure?
Tests whether AI agents can reverse engineer real-world, contamination-free binaries.
Which model leads the published ReverseEngBench snapshot?
GPT-5.6 Sol currently leads the published ReverseEngBench snapshot with 30.53% reverseengbench score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on ReverseEngBench?
5 AI models are included in BenchLM's mirrored ReverseEngBench snapshot, based on the public leaderboard captured on August 10, 2026.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 2,000+ readers.
One email each week. Unsubscribe anytime.