Author: AIImpish
-

Mistral Large 4 Goes Open Weight: A Trillion-Parameter Model You Can Download, With One License Catch
On October 29, 2026, Mistral AI released the weights of Mistral Large 4, nicknamed "Le Chonk." It is a 1-trillion-parameter, natively multimodal model that first appeared in public preview earlier this month. Anyone can now download it from Hugging Face and run it on their own hardware, but one group of businesses needs a separate…
-

Anthropic Expands AI for Science: Free Claude Credits Now Flow Through NSF, Sloan, Simons, and More
On October 28, 2026, Anthropic announced 20 new scientific foundations, institutions, and research projects in its AI for Science program, which gives free Claude credits to researchers on high-impact work. The shift: instead of only supporting individual scientists, Anthropic now routes credits through funders and major labs that can reach far more people. Coffee summary…
-

Canada Plans Up to $1.6 Billion Stake in Cohere: What “Sovereign AI” Means and Why It Matters
On October 29, 2026, Prime Minister Mark Carney announced that the Government of Canada will make an equity investment of up to $1.6 billion in Cohere, the Toronto-founded AI company. It is a commercial investment made alongside private backers including NVIDIA, Bell, and Schwarz Group, and it comes with conditions. Coffee summary FACT: Canada commits…
-

AWS Turns Step Functions Workflows Into AI Agent Tools: Let the Agent Decide, Let the Workflow Obey
AWS announced on October 29, 2026 that Step Functions workflows can now be used as tools by AI agents through Amazon Bedrock AgentCore Gateway. The agent decides when to run a workflow and what to send it; Step Functions then runs every step exactly as you defined it. It is a neat answer to the…
-

Claude Enterprise Adds Smart Reports and Pooled Budgets: Finally, a Way to See What AI Spend Buys
On October 28, 2026, Anthropic made three admin features generally available for Claude Enterprise: smart reports, in-app surveys, and pooled group budgets. Together they answer the question every manager paying for AI seats keeps asking: what are we actually getting for this money? Coffee summary FACT: Smart reports read a sample of a team's transcripts,…
-

Gemini Can Now Create and Edit Your Docs, PDFs, Slides, and Spreadsheets Right in the Chat
Google announced on October 29, 2026 that the Gemini app can now generate, preview, and edit files directly inside a conversation. You no longer have to copy text out of the chat to fix a document: you can open the file full screen, type edits yourself, or highlight a section and ask Gemini to change…
-

OpenAI Rewrites Its Model Spec Around Five Values: What Changes for ChatGPT Users
On October 29, 2026, OpenAI published a major update to its Model Spec, the public document that describes how its models are supposed to behave. Instead of a long list of rules and style tips, the Spec is now organized around five core values, and it calls the model "GPT" rather than "the assistant." Coffee…
-

OpenAI and ServiceNow Expand Their Partnership: Fixing IT Tickets From Inside ChatGPT
OpenAI and ServiceNow announced an expanded partnership on October 28, 2026, building on the deal they signed in January. The headline for everyday workers: IT and operations teams will be able to look up incidents, assess changes, and update ServiceNow records without leaving ChatGPT. Coffee summary FACT: A ServiceNow integration brings business context and workflow…
-

Google’s Connected Apps Platform: Any Business Can Now Apply to Live Inside Gemini and Search
On October 28, 2026, Google opened its Connected Apps Platform in public preview, letting U.S. developers build and submit apps that users can work with directly inside Gemini and Google Search. It could become the next kind of visibility for businesses: not just ranking in search results, but being the app the AI uses to…
-

NVIDIA Nemotron 3.5 Transcribe: An Open Model That Knows Who Said What, in 40 Languages
NVIDIA released Nemotron 3.5 Transcribe on October 28, 2026, an open-weight speech model that handles transcription, translation, speaker labels, and audio questions in one checkpoint. Alongside it, NVIDIA published Granary v2, the open speech dataset and pipeline used in its training. Coffee summary FACT: Transcribes 40 languages, translates 31 of them to or from English,…