AI Frontier · Aug 16, 2026
This roundup covers the past 7-day window, though the material is concentrated in Aug 10–13 (nothing new from Aug 14–16; see the note at the end). Ordered by publication time, most recent first.
Models and Capabilities
Gemini 3.7 Flash released (Google DeepMind · Aug 13)
Google DeepMind introduces Gemini 3.7 Flash, the latest iteration of the Gemini family optimized for speed- and cost-sensitive scenarios.
Source: https://deepmind.google/blog/introducing-gemini-3-7-flash/
The builder's guide to GPT-5.6 (OpenAI · Aug 13)
OpenAI publishes its GPT-5.6 builder's guide, covering how startup teams can build faster, more cost-efficient AI agents through smarter model selection and the new Responses API.
Source: https://openai.com/index/builders-guide-to-gpt-5-6
Ultrafast preview: GPT-5.6 Sol up to 14x faster (OpenAI · Aug 13)
OpenAI previews the Ultrafast API service tier, powered by Cerebras, which runs GPT-5.6 Sol at up to 14x the speed with output rates reaching 750 tokens per second.
Source: https://openai.com/index/previewing-ultrafast
Security-Related
Daybreak cybersecurity models expand: the cyber defense window is narrowing (OpenAI · Aug 10)
OpenAI releases GPT-5.6-Cyber, a dedicated cybersecurity model for authorized vulnerability research, exploit validation, and security testing, delivered through Daybreak Red.
Source: https://openai.com/index/expanding-daybreak-as-the-cyber-defense-window-narrows
Frontier cyber models opened to more trusted partners (OpenAI · Aug 10)
Approved Daybreak partners can now use OpenAI's frontier cybersecurity models to provide authorized, governed cybersecurity services to their customers.
Source: https://openai.com/index/putting-frontier-cyber-models-in-more-trusted-hands
Research and Other
Sign language AI in users' hands: the SL2T model (Google DeepMind · Aug 12)
DeepMind introduces a sign-language-to-text (SL2T) model, bringing new sign language recognition features to deaf and hard-of-hearing users.
Source: https://deepmind.google/blog/putting-sign-language-ai-into-users-hands/
AMIE medical AI system demonstrates real-time clinical video consultation (Google · Aug 11)
Google's research medical AI system AMIE demonstrated real-time clinical video consultation capability (in simulated scenarios) in a pioneering study.
Source: https://blog.google/innovation-and-ai/models-and-research/google-research/amie-video-consultations/
Enterprise AI adoption report: from assistance to execution (OpenAI · Aug 12)
OpenAI research shows enterprises accelerating adoption of agentic AI capable of autonomous execution, with leading firms continuing to widen their lead on deployment speed.
Source: https://openai.com/index/how-enterprises-put-ai-to-work
Known gap: the three ai_security sources — HiddenLayer Research, Lakera Blog, and Protect AI Blog — currently have no usable RSS/Atom feed and are permanently 0 rows in raw_items. This is expected and does not indicate a fetch failure. No new AI frontier items appeared in the Aug 14–16 window, likely a slowdown in vendor publishing cadence.