Mistral's Vibe does what ChatGPT, Claude, and Gemini charge for, and it's free

Published Jul 25, 2026, 10:30 AM EDT Nolen began their writing career in 2019, with three years dedicated to editing the Creative section at MakeUseOf. Their expertise lies at the crossroads of technology and creativity, covering areas like photography, video editing, and graphic design. Outside of work, you'll often find Nolen diving into a good book, writing their own stories, or playing video games. The big three chatbots are the big three for a reason. ChatGPT is still the default for most people. Claude has been my personal go-to for design and long-form work. Gemini's tie-in with Google's ecosystem is super useful when you already live in Docs and Drive. So none of them really beg to be replaced, they're genuinely good tools. The thing is, AI has gotten a lot bigger than those three names. I've been rotating through a fourth, lesser-known alternative, and it keeps pulling ahead on the features I actually need and use. It's Mistral's Vibe… Want to stay in the loop with the latest in AI? The XDA AI Insider newsletter drops weekly with deep dives, tool recommendations, and hands-on coverage you won't find anywhere else on the site. Subscribe by modifying your newsletter preferences! The chatbox is the least interesting part of Vibe Agents carry the whole experience Vibe is Mistral's chatbot, formerly known as Le Chat, and I've been keeping it in my rotation for a minute now primarily because it's one of the few powerful models that's not controlled by the US government, so there's less likely to be surprises like we had with Fable 5. In late May 2026, Mistral rebranded it and rebuilt it into an agent. It runs in three modes now, Chat, Work, and Code, and Code Mode is a whole developer setup with GitHub sandboxes and a VS Code extension that I'll skip for now because it's not relevant to my workflow. Chat mode is where the preloaded and custom agents live. It comes with a Data Analyst, Personal Tutor, Writing Assistant, Global Summarizer, or you can build your own with custom instructions, knowledge, and tools like web search or Canvas. I built a research agent for my UX study work with web search enabled and instructions to prioritize official primary sources. But it works for anything where resources are important. Work Mode is the one worth talking about though, and it's what turned Vibe into a proper agent product. You give it a task and it maps out the steps it intends to take and waits for your sign-off before proceeding. You can edit the plan, add clarifications, or push back if it's heading somewhere you don't like. Once it starts, every tool call is visible while it works, and it pauses to ask before sensitive actions like sending an email or modifying a file. ChatGPT's agent features and Claude's Cowork are paid-plan only. Vibe's Work Mode is on the free tier, which is hard to look past. It has a Canvas Useful for more than just designs and mockups Canvas is probably the tool I reach for most in Vibe. It's Mistral's version of Claude's Artifacts or ChatGPT's Canvas, basically a side panel where the model produces something you can edit in place instead of regenerating the whole response. You can use it for documents, code, mockups, structured data, and presentations that render as Marp slide decks you can preview inside the chat. The design-adjacent work is what I use it for most, but it's also great for putting together briefs and presentations. Vibe lets you have multiple Canvases open in one chat as tabs. So I can have a data table, a draft brief, and an SVG mockup all open at the same time and jump between them without losing my train of thought. There's proper version control too, with navigation arrows for moving between drafts, a change-tracking view that shows exactly what got edited, and a one-click restore. Comparing this to the other three kind of amplifies how well Mistral implemented it. Claude Artifacts are great for one-shot HTML, SVG, and React, but each one lives in its own bubble and there's no native multi-tab setup. ChatGPT Canvas is closest for the highlight-and-edit flow on text and code, but no multi-tab, Marp preview, or built-in mockup iteration. Gemini Canvas is fine for docs and code editing but leans on Google Slides handoffs, which breaks the whole flow. The one feature I didn't expect to use every week is on the free tier Scheduled tasks the big three charge you for Scheduled tasks weren't on my radar until I actually tried them, and now I have a few running in the background every week. The setup is a prompt, a cadence (once, daily, weekly, monthly, or yearly), and a time, and then Vibe runs it unattended and drops a notification when the result is ready. It's not just a saved prompt with a timer though: Scheduled tasks run in Work Mode, so they can use skills, connectors, web search, libraries, and projects natively. The task can actually go and do something, not just remind you it's Tuesday. I set up a couple just to test it at first and ended up keeping them because they're beyond useful. For example, one of the scheduled tasks is a weekly AI news digest that pulls the biggest announcements from the past week and drops a structured summary I can skim in five minutes. ChatGPT Tasks are paid-only, Gemini's Scheduled Actions are also paid-only, and Claude's scheduled tasks live inside Cowork, which needs a paid plan too. Everyone's watching the big three, meanwhile this one's been pulling ahead None of this makes ChatGPT, Claude, or Gemini worse than they were last month, and I'll probably keep my Claude subscription for now. It's more that Vibe has been picking off features I use constantly, putting them on the free tier, and rolling them out while the big three keep gating theirs behind Pro plans. I think if you've only been reaching for the household names, Vibe's worth an afternoon before you write it off. Mistral Vibe