Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-09-06 · high
One commit shipped across versions 4.0.60, 3.0.109 and 2.0.125 at once, so teams on older majors get the update without touching their package.json.
Why now: Vercel published all three releases on September 5.
On September 5, Vercel backported GPT-6 reasoning configuration into three major versions of its OpenAI provider package, per vercel/ai's release notes. The same commit, 17e489e, went into 4.0.60, 3.0.109 and 2.0.125, plus two Azure counterparts.
Most maintainers add a new provider feature to the current major and tell everyone else to upgrade. Vercel pushed the same commit into three live majors at once. A team still pinned to 2.0.x for compatibility reasons gets GPT-6 reasoning controls without changing a line of its own code.
Five package versions carrying one commit is more coordinated releasing than a single feature usually gets. Two of those versions are Azure builds of the same provider, so the OpenAI and Azure code paths for GPT-6 reasoning moved together.
The release notes don't say why Vercel chose to backport instead of pushing everyone to the newest major. They also don't say whether older majors will keep getting this kind of treatment. Whether it becomes standard practice for provider-level changes or was a one-off tied to GPT-6 is still open.
Each link below shares sources, entities, or timing with this story.
A spec is a press release until someone who didn't write it implements it. GitHub made Agent Plugins 1.0 generally available on August 12 across VS Code, Copilot CLI, the Copilot SDK, and the Copilot app on all plans. The spec, published August 6, was co-authored by AWS, Anysp...
The project positions against Cluely, Otter, Granola, and Fireflies with a Rust system-audio layer that works across Zoom, Teams, Meet, Slack, and Discord without per-app integrations. Transcripts, embeddings, and keys stay on the machine by default, with local RAG over past m...
Satya Nadella said companies routing everything through a single proprietary lab may not survive. His argument: you hand that lab your most sensitive business context, and the lab can turn it against you as a competitor. His prescription is an orchestration layer — keep the ha...
Eight thousand stars in a single day. That's what happened when Warp open-sourced its Rust-based, GPU-accelerated terminal on April 28. The repo shot to 47.9K total stars, making it the highest-velocity project on GitHub this week by a wide margin. But the interesting part isn...
OpenRouter and Vercel's AI Gateway both list $2.50/M input and $15.00/M output, down from $5.00/$30.00, while OpenAI's own API docs show standard pricing. Vercel's changelog dates it precisely: Aug 17 through Sep 18, all requests through AI Gateway. Gateway-only and time-boxed...
GPT-5.6 Luna went to $0.20 input / $1.20 output per million tokens on July 30. That's an 80% cut. Terra dropped 20%. Luna's input now undercuts Gemini 3.1 Flash-Lite ($0.25/$1.50) and sits at one-fifth of Claude Haiku 4.5's $1 input. Simon Willison covered the announcement and...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.