OpenAI held its first developer conference, DevDay, on November 6th. The announcements were substantial: GPT-4 Turbo with 128K context, the Assistants API, GPTs for custom agents, and significant price reductions.
GPT-4 Turbo and the price cut
GPT-4 Turbo is 3x cheaper for input tokens and 2x cheaper for output tokens compared to the previous GPT-4. With a 128K context window, it can process approximately 300 pages of text in a single API call. For applications that were doing multi-request chunking to stay within context limits, this changes the architecture. For applications where GPT-4 pricing made production economics difficult, the price cut changes the business case.
The Assistants API
The Assistants API provides built-in thread management, code interpreter execution, and file retrieval. Previously, building a multi-turn AI assistant required managing conversation history, building retrieval systems, and executing code in sandboxes, all as separate engineering work. The Assistants API abstracts those components into a managed service. The tradeoff is control: you lose visibility into the retrieval and code execution internals.
GPTs: custom agents without code
OpenAI launched GPTs, custom versions of ChatGPT that anyone can configure through a conversation interface. You describe what you want the GPT to do, give it instructions, upload knowledge files, and optionally connect it to APIs. The GPT Store was showed up as a marketplace for sharing GPTs. For enterprise users, this creates an accessible path from business users to customised AI tools without involving engineering.
What it signals for the AI application landscape
The pattern in developer conference announcements is moving the baseline capabilities up. Features that required significant engineering work in 2023, thread management, file retrieval, code execution, are now API features. This compresses the time from idea to working prototype. It also means the competitive advantage for AI application builders is moving up the stack, from who can stitch together the right APIs to who understands the domain problem deeply enough to build something genuinely useful.