Uploaded July 2026 | Updated September 2026, 2 weeks ago
Over the past year and a half, benchmarking has been Insilico's main focus in AI. We benchmark everything, from target ID and disease modeling to clinical trial outcome prediction. With 31 developmental candidates, we can now use our own programs as benchmarks. About 1,200 benchmarks per developmental candidate from start to nomination.
We introduced TargetPro, our new advanced target discovery model built on those benchmarks. Frontier LLMs like ChatGPT 5.5 and Opus 4.8 are about 10 to 20% worse than our best-in-class specialist tools. But when we post-train open source models like Qwen using MMAI Gem, they outperform those frontier models by more than 10 to 15%. Dramatically.
PandaOmics 6.0 is also out. Biggest update since launch. Full single cell data analysis, fully connected to target ID. This is exactly how we identified TNIK as the best target for IPF.
#InSilicoMedicine #Longevity #DrugDiscovery #AI #PandaOmics #Benchmarking #TargetDiscovery #GenerativeAI #LongevityResearch #Biotech #DrugDevelopment #AIforScience #Aging #LifeSciences #HongKong
Over the past year and a half, benchmarking has been Insilico's main focus in AI. We benchmark everything, from target ID and disease modeling to clinical trial outcome prediction. With 31 developmental candidates, we can now use our own programs as benchmarks. About 1,200 benchmarks per developmental candidate from start to nomination.
We introduced TargetPro, our new advanced target discovery model built on those benchmarks. Frontier LLMs like ChatGPT 5.5 and Opus 4.8 are about 10 to 20% worse than our best-in-class specialist tools. But when we post-train open source models like Qwen using MMAI Gem, they outperform those frontier models by more than 10 to 15%. Dramatically.
PandaOmics 6.0 is also out. Biggest update since launch. Full single cell data analysis, fully connected to target ID. This is exactly how we identified TNIK as the best target for IPF.
#InSilicoMedicine #Longevity #DrugDiscovery #AI #PandaOmics #Benchmarking #TargetDiscovery #GenerativeAI #LongevityResearch #Biotech #DrugDevelopment #AIforScience #Aging #LifeSciences #HongKong
