Vals BioMysteryBench agent leaderboard (BioMysteryBench (Vals))
We show this table for reference; we do not rank on it.
Independent agent runs on biological data investigations, split by human-solvable and human-difficult tasks.
BioMysteryBench score on BioMysteryBench (Vals) — September 27, 2026
We mirror the published biomysterybench score view for BioMysteryBench (Vals). Claude Sonnet 5.5 leads the public snapshot at 81.11%, followed by GPT-6 Astra (79.26%) and Claude Opus 5.5 (79.26%). We do not use these results to rank models overall.
Claude Sonnet 5.5
Anthropic
Terminus 2
Terminus 2
GPT-6 Astra
OpenAI
Terminus 2 · max reasoning
Terminus 2
Claude Opus 5.5
Anthropic
Terminus 2
Terminus 2
19 modelsAgenticCurrentDisplay onlyUpdated September 27, 2026
BioMysteryBench score table (19 models)
ScoreHow BioMysteryBench is shown here
BenchLM mirrors the public Vals AI BioMysteryBench leaderboard captured from https://www.vals.ai/benchmarks/biomysterybench and updated by Vals on September 27, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
BioMysteryBench 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.
Snapshot
The published BioMysteryBench (Vals) snapshot places Claude Sonnet 5.5 first at 81.11%. The third row is 1.85 points behind. The broader top-10 range is 11.85 points, so the table still separates the published systems.
19 models have been evaluated on BioMysteryBench (Vals). The benchmark falls in the Agentic category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. BioMysteryBench (Vals) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About BioMysteryBench (Vals)
Year
2026
Tasks
End-to-end bioinformatics investigations from raw data
Format
Average task accuracy across three runs
Difficulty
Agentic biological analysis
Vals runs Anthropic's biological-data tasks through a fixed Terminus 2 harness and reports overall, human-solvable, and human-difficult scores. Three full-benchmark runs and their uncertainty remain source-specific, display-only evidence.
Freshness and provenance
Version
BioMysteryBench (Vals) 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.
Questions
What does BioMysteryBench (Vals) measure?
Independent agent runs on biological data investigations, split by human-solvable and human-difficult tasks.
Which model leads the published BioMysteryBench (Vals) snapshot?
Claude Sonnet 5.5 currently leads the published BioMysteryBench (Vals) snapshot with 81.11% biomysterybench score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on BioMysteryBench (Vals)?
The September 27, 2026 snapshot contains 19 AI models.
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.