GPT-5.6 is a new family of three models: Sol, our new flagship model; Terra, a capable lower-cost option; and Luna, our fastest and most cost-efficient model. The safeguards we have built for this launch – our most robust yet – are built to deliver these models safely and at scale, around the world.

OpenAI lan Molecule.one nuduhake kepiye ahli kimia AI sing meh otonom nggunakake GPT-5.4 ningkatake reaksi kunci kanggo nggawe obat, saéngga majokake panaliten kimia medisinal.

Ngenalaké LifeSciBench, benchmark sing ditulis lan ditelaah ahli kanggo ngevaluasi cara sistem AI nangani tugas lan keputusan riset ilmu hayati nyata.

Kepiye output goblin nyebar ing model AI: linimasa, sebab utama, lan perbaikan ing balik keanehan sing didorong dening kapribaden ing prilaku GPT-5.

Sinaua carane Model Spec OpenAI dadi kerangka publik kanggo prilaku model, ngimbangi safety, kebebasan pangguna, lan akuntabilitas nalika sistem AI saya maju.

Cara OpenAI nggunakake pemantauan ranté pikiran kanggo nyinaoni misalignment ing agen coding internal—nganalisis deployment donya nyata kanggo ndeteksi risiko lan nguatake perlindungan keamanan AI.

IH-Challenge nglatih model supaya ngutamakake instruksi sing dipercaya, ningkatake hirarki instruksi, steerability keamanan, lan ketahanan marang serangan injeksi prompt.