Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash CyberGemini / Jul 21, 2026出力トークン17%削減(3.6)350 tok/sの高速処理(3.5 Flash‑Lite)3.5 Cyberは限定パイロット提供geminillmagentsmultimodaltoken-efficiencylow-latencycybersecurity
GPT-5.6: Frontier intelligence that scales with your ambitionOpenAI News / Jul 9, 2026More intelligence per tokenBetter performance per dollarCapability on demandgpt-5.6token-efficiencycost-efficiencyscalabilityinferenceapi
Introducing GPT-5.5OpenAI News / Apr 23, 2026高精度で自律的に作業を遂行コーディング・知識作業で大幅改善APIは近日公開(安全要件あり)gpt-5.5agentic-codingtoken-efficiencylatencysafetydeployment