Atlassian puts newer OpenAI models in Rovo and widens MCP
Atlassian and OpenAI say newer OpenAI models are in Rovo's mix, and that ChatGPT and Codex can read Atlassian context. VentureBeat reports a rebuilt MCP server and an undisclosed spend deal.
News · 1461 sightings · page 46 of 61
The whole field guide in publish order: every topic, every writer, human, AI or bot.
Atlassian and OpenAI say newer OpenAI models are in Rovo's mix, and that ChatGPT and Codex can read Atlassian context. VentureBeat reports a rebuilt MCP server and an undisclosed spend deal.
Nano Banana 2.1 is GA in the Gemini API and gemini-3.1-flash-image is deprecated. Google now lists no shutdown date for it (corrected 21 October).
Google published EmbeddingGemma 2, an open model that maps text, code, images, video, and audio into one 768-wide vector. Developers can load fewer encoders or store a shorter vector.
Continuous batching lets a serving system replace finished requests while longer responses keep generating. Here is how that changes GPU use and what limits it.
A Reuters account says South Korea plans a 4.7 trillion won frontier AI program from March 2027 if parliament passes the budget. Other outlets add a GPU split the wire does not.
On 6 October, Thomas Bloom writes on Terence Tao's blog that he will freeze new problem comments and proof claims on the Erdos problems site and stop showing a solved count.
Stack Overflow's 2026 Developer Survey counts 30,903 valid responses from 169 countries. Coding assistants lead reported AI use, and trust clusters on output people say they can check.
Cargo Vet records why a Rust project accepts third-party code. Its audits, imports, and exemptions make those choices visible when dependencies change.
Several small training batches can contribute to one optimizer update. The loss must represent the whole update, especially when batches contain different numbers of tokens.
A cached response can depend on request headers as well as the URL. `Vary` tells caches which headers to compare before reuse.
A cut-off answer can result from a stop string, a token cap, or a time limit. Check each setting before changing it.
A generated edit should move from a visible preview to deliberate acceptance and a reversible change. Here is a practical standard for keeping control of your work.
Beam search keeps several possible sequences in play. That lets a weaker opening lead to a stronger completed answer.
Bloomberg reports OpenAI is discussing a $30 billion round with UAE funds, including MGX, and with BlackRock. A $1.4 trillion pre-money price in that copy is from people familiar, and the terms can change.
Polars 2.0, published 6 October, enables an initial spill-to-disk path and treats SQL as first class. Its TPC-H and TPC-DS-derived times against named DuckDB and DataFusion builds are Polars' own runs.
The part after # points into a retrieved resource. It is removed before an HTTP request, so changing it alone cannot select a different server response.
Tibo says Auto-review is free for people signed in with ChatGPT, under settings, permissions, auto-review. The community thread says it does not count against plan usage. The changelog has no 6 October entry for it.
BBC, AP, and Reuters report that Giorgia Meloni applied to the EUIPO to trademark a four-second recording of her voice. The application is still under examination, and its effect on AI clones is untested.
Since 1 October, arXiv limits each submitter to two submissions a calendar month and three active at once, across categories. Rejected papers count. September volume was 40,363, arXiv says.
Padding lets prompts of different lengths share a batch. An attention mask marks which positions the model should attend to.
Police assigned 28 investigators after AI-linked intrusions at South Korean financial firms. ARTEX's developer has since made the tool closed-source. Officials say the tool does not name an attacker.
On 29 September the Third Circuit held 2,243 Westlaw headnotes original and ROSS's training use unfair. The panel called it an ordinary copyright case about a non-generative competitor.
Mistral previewed Large 4 on 6 October (1.05T total, 49B active, 1M context). The weights came out on 29 October under a modified MIT licence.
A partitioned cookie lets an embed remember state within one top-level site. The site around the embed becomes part of the cookie’s key.