Short Stack #13: Mistral released the weights for Mistral Large 4, its trillion-parameter model

Get 5 AI picks like these every weekday at 7am MT. Free, plus 50 vibe-coding prompts when you confirm.

Short Stack #13Fri, Oct 30, 2026

Image from huggingface.co: Mistral and Hugging Face logos announcing Le Chonk (Mistral Large 4)

Mistral released the weights for Mistral Large 4, its trillion-parameter model

Mistral published open weights for Mistral Large 4 ("Le Chonk") on Oct 29: a 1-trillion-parameter mixture-of-experts model that reads text and images. Mistral's claims: it ties the top open-weight model on Artificial Analysis' Coding Agent Index while using 47% fewer tokens per task, and it's built for defensive security work. API price is $0.70 in / $2.10 out per million tokens. The license is a modified MIT; hosted-model resellers need a commercial license.

⏱ 10-MIN TRY

Run a real bug-fix prompt through Mistral Large 4 on OpenRouter and compare the fix and token count with your usual model.

Source: mistral.ai

Image from blog.google: Gemini turning a table in a file into a graph

Gemini can now create and edit Docs, Sheets, slides and code files right in the chat

Google said on Oct 29 that the Gemini app can generate, preview and edit files inside the conversation. Type edits yourself or highlight a section and ask Gemini to change it. PDFs work on web and mobile; Google Docs, Sheets, slide decks, Markdown, TXT and code files are web only. It's available to all users.

⏱ 10-MIN TRY

Ask Gemini on the web for a one-page Markdown spec of a feature you're planning, then highlight the weakest section and ask it to rewrite only that part.

Source: blog.google

Image from canva.com: Introducing Canva Brush card with design tool icons

Canva Brush is a design model that edits your existing layouts

Canva launched Brush on Oct 29, a vision-language model that reads a design and makes the edits you ask for, from one line of copy to every slide in a deck. Results stay editable. Canva says it's up to 69% faster and 46% cheaper per edit than its previous design agent. It's live in Canva AI for Canva Pro, with wider access coming.

⏱ 10-MIN TRY

Open an old deck in Canva AI, pick Brush from the dropdown, and ask it to restyle every slide to one new color palette. Note anything it changed that you didn't ask for.

Source: canva.com

Reddit is moving behind a login as search traffic dries up

On its Oct 29 earnings call, Reddit said it will shift to a primarily logged-in experience, starting with tests outside the US this quarter. It said Google search referrals fell by more than 1 million a week on average in October as AI answers replace clicks. A logged-out front page and an incognito mode stay.

⏱ 10-MIN TRY

Check what share of your last 90 days of signups came from Reddit or Google. If it's big, pick one owned channel, like email, to grow.

Source: techcrunch.com

Illustration: a sound wave flowing through three connected pipeline blocks to a microphone

Deepgram bought Daily, the maker of the open-source Pipecat voice-agent framework

Deepgram announced on Oct 29 that it is acquiring Daily, maker of Pipecat, an open-source framework that wires speech-to-text, LLMs and text-to-speech into real-time voice agents. Terms weren't disclosed. Both companies say Pipecat stays open source and works with any vendor's models. Deepgram also launched RCS, an open protocol for passing audio cues, like a whisper, to the model.

⏱ 10-MIN TRY

Run Pipecat's quickstart with the speech and LLM providers you already use, and time how fast your bot answers.

Source: techcrunch.com

Mistral released the weights for Mistral Large 4, its trillion-parameter model

Gemini can now create and edit Docs, Sheets, slides and code files right in the chat

Canva Brush is a design model that edits your existing layouts

Reddit is moving behind a login as search traffic dries up

Deepgram bought Daily, the maker of the open-source Pipecat voice-agent framework

Get tomorrow's in your inbox.