Breaking
RU77 Killed, 120 Wounded in Israeli Air Strikes on Hezbollah Strongholds in Southern LebanonRUX Platform Declines to Remove Israeli Minister's Call for Lebanon's DestructionGLOBALNine critical after train crash near Bedford, one driver deadINTwo ASIs and Home Guard Suspended Over Assault and Sexual Abuse Allegations at Police StationUKFive injured in suspected Islamophobic attacks in EdinburghAUNetherlands Face Pressure Against Sweden After Japan DrawINFour PTI Leaders Jailed for Decade Over 2023 Riots; Ex-Foreign Minister Qureshi AcquittedINZelenskiy Returns Polish State Award Over Unit RenamingGLOBALHarry Kane's World Cup Start: A Stark Contrast to QatarINTLTrump Doubles Down on Criticism of Italian PM Meloni, Claims She Sought Photos for PopularityRU77 Killed, 120 Wounded in Israeli Air Strikes on Hezbollah Strongholds in Southern LebanonRUX Platform Declines to Remove Israeli Minister's Call for Lebanon's DestructionGLOBALNine critical after train crash near Bedford, one driver deadINTwo ASIs and Home Guard Suspended Over Assault and Sexual Abuse Allegations at Police StationUKFive injured in suspected Islamophobic attacks in EdinburghAUNetherlands Face Pressure Against Sweden After Japan DrawINFour PTI Leaders Jailed for Decade Over 2023 Riots; Ex-Foreign Minister Qureshi AcquittedINZelenskiy Returns Polish State Award Over Unit RenamingGLOBALHarry Kane's World Cup Start: A Stark Contrast to QatarINTLTrump Doubles Down on Criticism of Italian PM Meloni, Claims She Sought Photos for Popularity
Newsgather

gpt-5.4

Steady6 stories3 sourcesLast updated: 04.05.2026

Latest Stories

23-Jähriger löst mit GPT-5.4 Pro jahrzehntealtes Erdős-Problem #1196
Developing
Science·4/29/2026AI summary

23-Jähriger löst mit GPT-5.4 Pro jahrzehntealtes Erdős-Problem #1196

Der 23-jährige Liam Price hat mit GPT-5.4 Pro das Erdős-Problem #1196 gelöst, das sich mit der Erdős-Summe primitiver Mengen befasst. Die KI bewies, dass die Summe asymptotisch gegen 1 konvergiert, und verbesserte damit die bisher beste Schranke von Jared Lichtman erheblich. Der Beweis nutzt Markov-Ketten zur Modellierung der Primfaktor-Struktur ganzer Zahlen – ein neuartiger Ansatz, den Terence Tao als engere Verbindung zwischen analytischer Zahlentheorie und stochastischer Prozesstheorie lobte. Das Problem wurde formal im Beweissassistenten Lean verifiziert.

H
Heise Online
OpenAI Announces GPT-5.5, Its 'Smartest and Most Intuitive' Model Yet
Developing
Tech·4/23/2026AI summary

OpenAI Announces GPT-5.5, Its 'Smartest and Most Intuitive' Model Yet

OpenAI announced GPT-5.5, describing it as its smartest and most intuitive model yet, featuring improved code writing and debugging, online research capabilities, and cross-tool functionality with enhanced safeguards. The release comes amid intensifying competition with Anthropic and just before Elon Musk's high-profile trial against OpenAI executives beginning Monday in Oakland, California.

T
The Verge