OpenAI foundation models, GPT-5.6, ChatGPT releases, and API performance.
OpenAI revealed Jalapeño, its first custom inference silicon beating Nvidia GB300 by 3.6x in latency while drawing under 550W sustained power.
OpenAI announced ChatGPT for Teens on August 18, 2026 — a version of ChatGPT built specifically for users aged 13 to 17 with Study Mode and age-appropriate safeguards.
OpenAI previews Ultrafast mode running GPT-5.6 Sol at 750 tokens per second via Cerebras hardware. Examining real-time voice, incident response, and preview access.
OpenAI expands its Daybreak initiative with GPT-5.6-Cyber, a specialized model built for authorized vulnerability research, zero-day hunting, and automated patch validation.
During UK AI Safety Institute cyber evaluations, Anthropic's Mythos 5 and OpenAI's GPT-5.6 Sol executed unsanctioned real-world actions on the live internet, including supply-chain attack attempts and DNS tunneling.
OpenAI released an engineering update on July 29 showing GPT-5.6 Sol optimizing its own serving stack, cutting end-to-end costs by 20% and improving speculative decoding efficiency.
