The week's real tension is cheap access versus production control
Veo 3.1, and The Deploy Log for this week. Every price and claim read off the publisher's own page, with a call on each.
Google DeepMind
Veo 3.1
DEPLOYif you make short-form video
WHAT SHIPPEDVeo 3.1 Ingredients to Video now supports native vertical 9:16 output, richer dialogue, consistent characters and backgrounds, and upscaling to 1080p and 4K.
THE CATCHThe update is available in the Gemini app, YouTube Shorts, Flow, the Gemini API, Vertex AI, and Google Vids, but 1080p and 4K options are only on Flow, the API, and Vertex AI.
IF YOU RUNa YouTube Shorts channel, you can now generate full-screen vertical storytelling without cropping or quality loss
SOURCEGoogle DeepMind, Veo 3.1 Ingredients to Video: More consistency, creativity and control
The Deploy Log
Models and APIs
| DEPLOYif you run long coding tasks GPT 5.2 Codex. GPT 5.2 Codex is now available on Vercel AI Gateway with no other provider accounts required, combining GPT 5.2's professional knowledge work with GPT 5.1 Codex Max's agentic coding. Vercel |
Tools and products
| DEPLOYif you build on Vercel with AWS data AWS databases on Vercel. Developers can provision Aurora PostgreSQL, Amazon DynamoDB, and Aurora DSQL directly from the Vercel dashboard, and new AWS teams get a free starter plan with $100 in credits. Vercel |
| DEPLOYif you switch models across providers OpenResponses API on AI Gateway. Vercel AI Gateway is a day 0 launch partner for the OpenResponses API, supporting text generation, streaming, tool calling, image input, reasoning, and provider fallbacks. Vercel |
| DEPLOYif your models lack native search Perplexity web search. Provider-agnostic web search on AI Gateway lets any model search the web in real-time with a single line of code, charged at $5 per 1,000 requests. Vercel |
Left out
- ChatGPT Go: a figure in the draft could not be found on the publisher's own page
- MedGemma 1.5 4B: the publisher's page does not show it available yet
- Sandbox CLI SSH: the publisher's page does not show it available yet
- Reduced build times: a figure in the draft could not be found on the publisher's own page
- Team-wide Deployment Protection defaults: the publisher's page does not show it available yet
- Protection bypass multiple secrets: it did not carry a number that moved
- Project-level deployment suffixes: it did not carry a number that moved
- Docs pages as Markdown: it did not carry a number that moved
- AI Elements voice components: it did not carry a number that moved
- Vercel Agent code reviews: it did not carry a number that moved
- Hono and Express build mode: it did not carry a number that moved
The pattern
Google DeepMind gives Veo 3.1 native 9:16 output and 4K upscaling, but reserves 1080p and 4K for Flow, the API, and Vertex AI, leaving the Gemini app and YouTube Shorts on lower tiers. Vercel opens AWS databases and GPT 5.2 Codex through one dashboard and gateway, then charges $5 per 1,000 Perplexity web search requests. The week's real tension is cheap access versus production control: Veo 3.1 makes vertical video free to start, while Vercel converts $100 in AWS credits, OpenResponses API support, and provider fallbacks into a metered funnel where Aurora PostgreSQL, DynamoDB, Aurora DSQL, and GPT 5.1 Codex Max sit behind the same gateway as a $5 search toll.
No single item says this, but together they show a split between creative output and infrastructure billing. Veo 3.1 pushes richer dialogue and consistent characters into YouTube Shorts without cropping, yet withholds 4K from that surface. Vercel's day 0 OpenResponses launch and provider-agnostic search make any model capable of real-time web access, but the $5 per 1,000 requests price and the $100 starter plan frame that capability as a paid tier. The pattern is not that AI is becoming cheaper; it is that the cheap layer is the demo, and the production layer is the toll.