# BioSecBench-RespiratoryVirus-Surveillance results > Tests whether AI agents can analyze laboratory results, genetic data, and health records to identify respiratory viruses, check how well diagnostic tests work, and track how these viruses change and spread. These capabilities support public-health efforts to spot outbreaks and estimate how many people are getting sick. Scores show the share of attempts that reach the correct answer, counting refusals and errors as failures. A higher score means the model is more accurate at these tasks. Data updated: 2026-11-05 Artifact generated: 2026-11-05T18:10:31.683Z - Domain: Biosecurity - Horizon: short - Assessment: Capabilities - Published score scope: Current V1 evaluation set (110 evaluations) Scores use a 0–100 percentage scale. Consult the repository and paper for benchmark-specific task construction, grading, aggregation, and uncertainty methodology. Results compare model-and-harness combinations. LSVP: Anthropic's safety-filter-exempt evaluation variant, which produces fewer refusals on biology prompts. LSVP results are reported alongside the model's standard run and are not a replacement for it. | Rank | Model | Harness | Provider | Score | Variant | | ---: | --- | --- | --- | ---: | --- | | 1 | Claude Opus 5.5* | Pi | Anthropic | 36.4% | LSVP | | 2 | Claude Opus 5.5* | Claude Code | Anthropic | 31.2% | LSVP | | 3 | Claude Opus 5.5 | Pi | Anthropic | 25.4% | | | 4 | Claude Sonnet 5.5* | Pi | Anthropic | 24.2% | LSVP | | 5 | Claude Fable 5.1* | Pi | Anthropic | 22.7% | LSVP | | 6 | Claude Sonnet 5.5* | Claude Code | Anthropic | 21.5% | LSVP | | 7 | Claude Opus 5.5 | Claude Code | Anthropic | 19.4% | | | 8 | Claude Fable 5.1* | Claude Code | Anthropic | 18.5% | LSVP | | 9 | Claude Opus 5* | Pi | Anthropic | 18.5% | LSVP | | 10 | Claude Opus 5* | Claude Code | Anthropic | 17.9% | LSVP | | 11 | Gemini 4 Argon | Pi | Google | 17.6% | | | 12 | Claude Sonnet 5.5 | Pi | Anthropic | 17.3% | | | 13 | Grok 4.7 | Pi | SpaceXAI | 15.4% | | | 14 | Claude Opus 5 | Pi | Anthropic | 14.6% | | | 15 | Claude Sonnet 5.5 | Claude Code | Anthropic | 14.2% | | | 16 | Grok 4.7 | Grok Build | SpaceXAI | 13.6% | | | 17 | Claude Opus 5 | Claude Code | Anthropic | 13.3% | | | 18 | GPT-6 Astra | Pi | OpenAI | 13.3% | | | 19 | GPT-6.1 Sol | Codex | OpenAI | 12.4% | | | 20 | GPT-6.1 Sol | Pi | OpenAI | 12.1% | | | 21 | Grok 4.6 | Pi | SpaceXAI | 11.5% | | | 22 | Grok 4.6 | Grok Build | SpaceXAI | 11.5% | | | 23 | GPT-6.1 Terra | Codex | OpenAI | 11.2% | | | 24 | GPT-6 Astra | Codex | OpenAI | 10.3% | | | 25 | GLM 5.3 | Pi | Z.ai | 7.9% | | | 26 | Kimi K3 | Pi | Moonshot | 7.6% | | | 27 | Gemini 3.8 Flash | Pi | Google | 7.3% | | | 28 | Nemotron 3 Ultra 550B A55B | Pi | NVIDIA | 7.3% | | | 29 | GPT-6.1 Terra | Pi | OpenAI | 6.7% | | | 30 | GPT-6 Luna | Pi | OpenAI | 6.1% | | | 31 | Mistral Large 4.0 | Pi | Mistral | 6.1% | | | 32 | GPT-5.6 Sol | Pi | OpenAI | 5.8% | | | 33 | GPT-5.6 Sol | Codex | OpenAI | 5.8% | | | 34 | GPT-5.6 Terra | Codex | OpenAI | 5.8% | | | 35 | DeepSeek V4.1 Flash | Pi | DeepSeek | 5.2% | | | 36 | GPT-6 Luna | Codex | OpenAI | 5.2% | | | 37 | GPT-5.6 Terra | Pi | OpenAI | 4.8% | | | 38 | GPT-5.6 Luna | Pi | OpenAI | 4.5% | | | 39 | Nemotron 3.5 Lightning | Pi | NVIDIA | 3.9% | | | 40 | Claude Fable 5.1 | Pi | Anthropic | 3.6% | | | 41 | Claude Fable 5.1 | Claude Code | Anthropic | 3.6% | | | 42 | GPT-5.6 Luna | Codex | OpenAI | 3.6% | | ## Sources and downloads - [Paper PDF](/papers/respvirus-surveillance.pdf) - [Original publication](https://latch.bio/respvirus-surveillance) - [Repository](https://github.com/latchbio/biosecbench-respvirus-surveillance) - [Interactive benchmark](https://benchmarks.bio/respvirus-surveillance) - [JSON](https://benchmarks.bio/benchmarks/biosecbench-respvirus-surveillance/latest.json) - [CSV](https://benchmarks.bio/benchmarks/biosecbench-respvirus-surveillance/latest.csv) - [Current V1 cross-benchmark leaderboard](https://benchmarks.bio/results/v1/)