Dec 18, 2025
Faster Models, App Stores, and AI Upskilling
š§© The Gist
Google introduced Gemini 3 Flash, described as frontier intelligence built for speed at a fraction of the cost, with benchmarks and global availability noted. OpenAI opened an ināproduct app directory and submission flow for ChatGPT, plus launched an Academy to help newsrooms use AI responsibly. Greptile published a crossāindustry State of AI Coding 2025 study on engineering velocity, adoption, and benchmarks. The week also featured an industry reality check from AWSās CEO on junior developers, a Coursera and Udemy combination aimed at AIāera skills, and fresh hiring in AI filmmaking tools.
š Key Highlights
- Google announced Gemini 3 Flash, positioned as frontier intelligence built for speed and lower cost, with benchmarks and global availability referenced.
- OpenAI now accepts developer app submissions to ChatGPT, with approved apps listed in a new directory, updated guidelines, and an Apps SDK for chatānative experiences that can perform realāworld actions.
- Greptileās State of AI Coding 2025 is a crossāindustry study covering engineering team velocity, AI tool adoption, model growth trends, and performance benchmarks.
- AWS CEO Matt Garman argued AI cannot replace junior developers, calling the idea one of the dumbest and outlining reasons it falls short.
- OpenAI launched the Academy for News Organizations with the American Journalism Project and The Lenfest Institute, offering training, practical use cases, and responsibleāuse guidance.
- Coursera announced plans to combine with Udemy to empower the global workforce with skills for the AI era.
- YCābacked Flick is hiring a founding frontend engineer to build a Figmaālike interface for AI filmmaking.
šÆ Strategic Takeaways
- Models and performance
- Emphasis on speed and cost efficiency signals a push to make frontierālevel intelligence more accessible to mainstream workloads.
- Developer ecosystem
- An app submission pipeline and directory inside ChatGPT creates a clearer path for discovery and distribution of chatānative apps.
- Workforce and education
- Leaders are pushing against AIāonly staffing strategies for engineering, while major learning platforms and newsroom programs focus on structured upskilling.
- Creative tooling
- Hiring and product work in AI filmmaking point to expanding vertical tools for media creation.
š§ Worth Reading
- The State of AI Coding 2025 (Greptile): A crossāindustry report on AI software development that examines team velocity, tool adoption, model trends, and benchmarks. Practical takeaway, teams can use its benchmarked themes to evaluate where AI code review and coding assistants fit in their workflows and where performance gains are most likely.
Introducing GPT-5.2-Codex (openai.com) GPT-5.2-Codex is OpenAIās most advanced coding model, offering long-horizon reasoning, large-scale code transformations, and enhanced cybersecurity capabilities. openai
Addendum to GPT-5.2 System Card: GPT-5.2-Codex (openai.com) This system card outlines the comprehensive safety measures implemented for GPTā5.2-Codex. It details both model-level mitigations, such as specialized safety training for harmful tasks and prompt inj⦠openai
Introducing GPT-5.2-Codex (openai.com) GPT-5.2-Codex is OpenAIās most advanced coding model, offering long-horizon reasoning, large-scale code transformations, and enhanced cybersecurity capabilities. openai
Deepening our collaboration with the U.S. Department of Energy (openai.com) OpenAI and the U.S. Department of Energy have signed a memorandum of understanding to deepen collaboration on AI and advanced computing in support of scientific discovery. The agreement builds on ongo⦠openai
Evaluating chain-of-thought monitorability (openai.com) OpenAI introduces a new framework and evaluation suite for chain-of-thought monitorability, covering 13 evaluations across 24 environments. Our findings show that monitoring a modelās internal reasoni⦠openai
Updating our Model Spec with teen protections (openai.com) OpenAI is updating its Model Spec with new Under-18 Principles that define how ChatGPT should support teens with safe, age-appropriate guidance grounded in developmental science. The update strengthen⦠openai