Insilico Bench | Insilico Medicine

Insilico's Proprietary AI Benchmarks

Proprietary benchmarks developed by Insilico Medicine to evaluate AI systems across the drug discovery pipeline — from target identification to clinical development. A research initiative by Insilico Medicine.

Insilico Bench Benchmarks Explorer

Interactive exploration of benchmarks across Biology, Chemistry, Materials, and Longevity. Browse and compare model performance with built-in visualizations.

Models Ranked

16 of 17 total Models Ranked
198 of 203 total Benchmarks
5 active Categories

Top Model: GPT 5.5
Insilico score: 66.4

Included benchmarks (198)

Aging Prediction: Balanced accuracy [0, 1]
GTEx Expression: accuracy_score [0, 1]
GTEx Pairwise Binary: Balanced accuracy [0, 1]
GTEx Pairwise Ternary: accuracy_score [0, 1]
Longevity Synergy (Full): Balanced accuracy [0, 1]
Longevity Synergy (Minimal): Balanced accuracy [0, 1]
Methylation Age Choice: accuracy_score [0, 1]
Methylation Age Pairwise: Balanced accuracy [0, 1]
Methylation Age Regression: Spearman correlation [-1, 1]
NHANES Mortality Classification: Balanced accuracy [0, 1]
NHANES Pairwise Comparison: Balanced accuracy [0, 1]
NHANES Time-to-Event: accuracy_score [0, 1]
NHANES TTE Regression: Spearman correlation [-1, 1]
Olink Pairwise Comparison: Balanced accuracy [0, 1]
Olink Protein Classification: Balanced accuracy [0, 1]
TCGA Survival Prediction: Balanced accuracy [0, 1]
...

Ranked Table

#	Model	Insilico Score	Biology	Affinity and Binding	Chemical Synthesis	ADMET, PK & Safety	Clinical Trials
1	GPT 5.5	66.4	35	71	30	70	81
2	Claude Opus 4.8	58.6	30	57	26	69	78
3	Claude Opus 4.6	58.5	59	62	9	63	72
4	Claude Opus 4.5	57.9	63	59	9	66	73
5	Claude Opus 4.7	56.9	—	70	12	70	—
6	DeepSeek v3.2	56.3	56	62	1	59	68
7	Claude Sonnet 4.5	55.9	61	62	6	62	59
8	GPT 5.2	55.1	57	62	4	61	53
9	Grok 4.1	55.0	51	61	14	59	60
10	Kimi K2.5	52.8	58	56	9	61	64
...	...	...	...	...	...	...	...

Insilico's Proprietary AI Evaluation Suite

Insilico Bench is Insilico Medicine's proprietary benchmarking platform, built to rigorously evaluate AI models on real-world drug discovery tasks using internal datasets and domain expertise.

Proprietary Data

Benchmarks built on Insilico's internal drug discovery datasets and pipelines.

LLM Evaluation

Comprehensive testing of large language models on domain-specific tasks.

Transparent Metrics

Clear, reproducible evaluation metrics with public leaderboards.

Industry-Proven

Grounded in Insilico Medicine's track record of AI-driven drug candidates in clinical trials.