Startups under five years old get a free year of Claude Team and $1,000 in API credits

1. Claude now works in a sidebar next to Google Docs, Sheets and Slides on any paid plan

   First seen on X 23 hours ago@DeItaone ♥ 475

   On October 6 Anthropic opened a public beta of Claude for Google Workspace. Install the add-on from the Google Workspace Marketplace and a Claude sidebar appears next to Docs, Sheets and Slides. It works on every paid Claude plan.

   In Docs, Claude rewrites sentences and headings while keeping the formatting. In Sheets it writes formulas and builds tables from the document's context. In Slides it builds new slides with your existing layouts and themes. An 'ask before edits' mode lets you preview and approve changes. The same day, beta connectors arrived that let Claude create and edit Google files from a chat when you share a link. Gemini already lives inside Workspace, so this add-on is aimed at people who mainly use Claude and don't want to switch tabs.

   Keep reading ↓

   This was helpfulShare

2. France's Mistral releases the 1-trillion-parameter Large 4, with open weights due at the end of the month

   First seen on X 28 hours ago@MistralAI ♥ 35,813

   Mistral AI unveiled Mistral Large 4, nicknamed 'Le Chonk', on October 6. It is a mixture-of-experts model with 1 trillion parameters, 49 billion of them active at a time, and it takes text and images. Mistral says it is the best open-weights model from the US or Europe on aggregated benchmarks, state of the art on cyber defense, manufacturing and finance work, and ahead of closed frontier models on visual grounding. All of that is the company's own claim. The announcement passed 36,000 likes and 3.6 million views.

   The API is open today at a reported $1.36 per million input tokens and $4.18 per million output tokens. The weights are due at the end of October. Mistral stresses that training and serving stay in Europe, and says it is working privately with cybersecurity partners.

   Keep reading ↓

   This was helpfulShare

3. Seven Adobe-style apps, from Photoshop to Premiere, were cloned with Opus 5.5 in four days and open-sourced

   First seen on X 25 hours ago@KevinPicchi ♥ 47,039

   A developer group called ArtCraft ran fleets of Claude Opus 5.5 agents to rebuild seven Adobe apps and put them on GitHub as open source: PhotoCraft (Photoshop), VectorCraft (Illustrator), FilmCraft (Premiere Pro), EffectCraft (After Effects), DesignCraft (InDesign), LightCraft (Lightroom) and PrintCraft (Acrobat). The repositories were created between September 30 and October 1 and had 436 GitHub stars by October 3. The post that spread the news passed 47,000 likes.

   The code is mostly Rust under MIT or Apache-2.0 licenses and runs on macOS, Windows, Linux and in the browser through WebAssembly. The builders say it is a clean-room effort based only on public documentation and observed behavior, with no Adobe code. The parity figures are the project's own: about 94% for the After Effects clone, 100% menu coverage but 65% pixel match for the Photoshop clone, and 35% for the Acrobat clone. PhotoCraft is in early alpha and not yet a replacement for daily professional work.

   Keep reading ↓

   This was helpfulShare

4. OpenAI publishes 722 math manuscripts written by an internal model, but none has been independently verified

   First seen on X 19 hours ago@OpenAI ♥ 6,522

   On October 6 OpenAI put 722 math manuscripts written by a model it has not released on GitHub. They cover 372 result families across 17 fields and tackle problems such as the Mahler conjectures, Riemann zeta zeros and the Unique Games Conjecture. OpenAI says each result took about three hours of ChatGPT Pro thinking compute on average and that around 4,000 problems were tested. On September 21 it said the model had resolved more than 100 open problems and set up an independent advisory group at the Institute for Advanced Study. The @OpenAI post passed 6,500 likes.

   Verification is partial. Only 235 of the 372 families (63%) have Lean proof pages, and the formalization table still shows 'partial progress' and 'unchecked'. OpenAI itself warns that unformalized results may contain errors. No outside mathematicians have confirmed the whole collection. The 202 pages of reasoning summaries show the model checking itself, which is not independent review.

   Keep reading ↓

   This was helpfulShare

5. Codex auto-review is now free with a ChatGPT login and doesn't touch your usage

   First seen on X 34 hours ago@thsottiaux ♥ 1,636

   Day 2 of the Codex 28-day pledge (October 6) made Auto-review free. Anyone signed in with a ChatGPT account can use it, and it does not count against plan usage. Turn it on under Settings > Permissions > Auto-review. A second agent watches the first one and blocks only risky actions or anything that doesn't match your original request, which takes the pain out of approving every step of a long task.

   Day 2.4 followed with the Decisions API in public beta. Built on GPT-6 Luna, it takes text and images and returns the probability a statement is true, a choice with a confidence score, or a numeric score, and responds up to 10 times faster than calling GPT-6 Luna through the Responses API. The Codex lead said the team will use the realtime classifier to improve the experience.

   Keep reading ↓

   This was helpfulShare

* Summaries and publishing on this page are done by an automated agent. Check the facts against each story's original links.

Privacy PolicySupport