# Vibe by Mistral (ex-Le Chat) — user reviews > 676 indexed App Store user reviews of Vibe by Mistral (ex by Mistral AI, rated 4.80/5 from 1,229 ratings. - Source: https://appshunter.io/ios/app/vibe-by-mistral-ex-le-chat/id6740410176/reviews - App overview in markdown: https://appshunter.io/ios/app/vibe-by-mistral-ex-le-chat/id6740410176.md - Rating breakdown: 5★ 1,073 · 4★ 88 · 3★ 28 · 2★ 10 · 1★ 30 ## What do users think of Vibe by Mistral (ex? Overall sentiment: positive **Pros** - 🚀 Lightning-fast responses - 👍 User-friendly design - 🔒 Privacy-focused - 💡 Innovative AI models - 🤖 Powerful and accurate - 🆓 Free to use **Cons** - 🐛 Occasional bugs and crashes - 🔍 Limited features compared to web version - 🤖 Short or incomplete responses - 🤔 Inconsistent performance **Commonly reported bugs** - Keyboard issues obscuring UI elements - Inconsistent app behavior on different devices - Frequent network errors **Most requested features** - Add voice input and output - Implement system theme support - Improve accessibility for screen readers - Expand file attachment capabilities ## Newest reviews (20 of 676) ### 3/5 — Great models let down by the app *2026-08-18* This app really needs to allow connection to custom MCP servers in order to allow users more flexibility. Currently I can only connect to predefined ones, however, I have a bunch of streaming http endpoints I’d like to hit that would fix my workflow here. **Developer response:** Thank you for your feedback. We understand the need for custom MCP server connections. For assistance, please visit our help center at http://help.mistral.ai/. ### 3/5 — Can’t code properly *2026-08-18* I’m trying to grasp what Mistral chat can do so I can decide if it’s a viable possibility to use it. They improved mermaid views but the model fails to understand some commands. Simple stuff works, complex ideas do do not materialize and no recover **Developer response:** We understand you're facing challenges with coding functionality and command understanding. For further assistance, please visit our help center at http://help.mistral.ai/. ### 3/5 — New updates ruined the app *2026-08-11, version 2.8.0* I love mistral for research and its speed but these updates have ruined the experience. ### 4/5 — Not bad, still not usable to justify paid tier *2026-08-10* I have been experimenting with basic Canvas and dashboard creation… seems crude and it takes lots of iterations to get something useful. But it works eventually I think **Developer response:** Creating something useful can take time, and we appreciate your feedback. Your insights on canvas and dashboard creation will help us improve usability. Thank you. ### 3/5 — Liquid Glass *2026-07-18, version 2.5.0* I really enjoy this service and the app overall BUT Liquid Glass design means buttons float over the scrolling text and it’s a mess. Look, I know it’s Apple’s fault for coming out with this “design” language but please have some way to turn it off. Respect the system-wide “Reduce Transparency” setting so it displays a toolbar rather than floating buttons over faded text. It’s just so ugly and really hard to decipher the UI. I wish Apple never encouraged this. Please fix this somehow. Thank you. ### 3/5 — New updates ruined the app *2026-07-14, version 2.5.0* I love mistral for research and its speed but these updates have ruined the experience. ### 5/5 — Tolle App mit Ausbaupotential *2026-07-13* Super App für Anfragen, Recherchen oder Analysen. Benutze ich mittlerweile seit Monaten und täglich. Schön zu sehen, dass es auch europäische und konkurrenzfähige Produkte gibt. Es gibt aber noch Ausbaustufen. Beispielsweise die Bildgenerierung ist aktuell mehr ein Spaßfeature, ernsthafte Ergebnisse kommen hier nicht. Selten vorkommend, aber es gibt verworrene Situationen, wenn die KI nicht weiß welcher Tag heute ist und ein Datum von gestern zeigt. Ich würde mir auch sehr eine CarPlay App wünschen. Sowas könnte man auch sehr schlank aufbauen. Trotzdem super App und ich freue mich auf die weitere Entwicklung. **Developer response:** Es freut uns, dass die App Ihnen bei Anfragen und Recherchen hilft. Wir schätzen Ihr Feedback zur Bildgenerierung und den Unstimmigkeiten. Ihre Idee für eine CarPlay-App nehmen wir gerne auf. Vielen Dank! ### 3/5 — Great AI, but missing a basic feature *2026-07-04* 5 stars for the AI itself. it’s fantastic. But as a Pro user, I’m shocked that I can’t use my custom Voxtral voices in the Vibe app. I went through the effort of cloning a voice, only to find out it’s useless on mobile. This is a no-brainer integration that would make the experience 10x better. Fix this, and I’ll happily update my review to 5 stars. It deserves 5 stars alone for still being supported on an Iphone 8 running iOS 16. Thank you for that genuinely. Also, why can’t agents have their own voices? If I have 5 agents, and they all sound the same, it's reductive. This would make the experience SO much better. Please add voice customization per agent! Not just for this app but in general for the Mistral Ecosystem. ### 5/5 — Bad name, Good LLM system, much better than ChatGPT, Gemini *2026-07-04, version 2.5.0* Vibe is a bad Gen Z name. Vibe is a verb. Vibe should not be a noun. Mistral.AI’s chatbot costs less than more prominent generative systems, while doing everything they do. I rarely hit a usage limit, glitch, or slowdown, despite using their app and website many times a day, often in long conversations. The app and website could improve the input area’s clarity, especially when uploading files alongside text. The interface feels cluttered in these moments. Additionally, an option to remove the voice-to-text icon would be welcome, as I prefer using Siri for local dictation to save cellular bandwidth. The new “work mode” should be dynamically integrated with the “chat mode,” so users don’t need to pick one at the start. Unifying the two modes would also make searching for a past conversation easier to find. But overall it is a well-built system capable of image generation, image processing, file analysis, data parsing, current events research, and summarizing immense amounts of text (like multi-hour podcast transcripts or digging through literature). A good app, a very good value. ** I read Mistral’s reply to this review and I will leave this review until the cited challenges are solved. ### 5/5 — Great alternative from the main stream models. *2026-07-03* Been using local Mistral models for quite a while now and I’ve always been pleased with the results. I finally discovered the Mistral app, and I’m genuinely impressed with the output it fits my workflow really well. My only concern is that LLM companies start out strong like this, but as investors and larger organisations begin to influence their direction, the models gradually stop delivering the experience that users originally fell in love with. **Developer response:** Thank you for your feedback. We appreciate your insights on quality. ### 5/5 — First choice AI *2026-07-03, version 2.5.0* Mistral is easy to use, fast and accurate. It has quickly become by first choice AI. ### 5/5 — exceptionally capable model *2026-06-30, version 2.4.0* Amazing the greatest value I’ve had ### 3/5 — Excellent premier contact *2026-06-29* Je vais passer le mot ici au Québec ### 2/5 — It’s Confused. I’m not. *2026-06-29* I’ll make this clear. It’s been a while since I’ve encountered a such a convoluted product. There seems to be no central tenet. It almost seems unaware of its own function and benefits. The barriers to entry started with trying to figure out how to even sign up for an account. I’d say it contradicts itself in terms of what it can and can’t do, but no. It appears to have been designed in the ‘I can do this but I can’t do that, unless I’m doing that other thing. Then I can, but when there I can’t do the first other thing that I was never sure about in the first place’. If that sentence twists your head around your neck 360 that will give you an idea of the bizarre confusing journey I had just trying to get it to explain itself and its features; which frankly it sadly failed to do. Clearly a powerful app and I’d like to say purpose specific but I cannot because it could not coherently nor consistently inform me of it’s CANS. It did however spend plenty of time listing its CANT’S. The usage restrictions, by the way are INSANE. Very hard to make a decision to go pro when it says times up before Usain Bolt’s hit 100 metres. I would like to provide a hypothetical equivalence. “Hello I would like to buy this car” - “Great” - “May I test drive it?” - “No”. Fin. **Developer response:** Navigating a new app should be smooth, and we acknowledge the challenges you faced. Your feedback on the sign-up process and feature clarity is valuable. We appreciate your insights and will consider them for future improvements. For assistance, please visit our help center at http://help.mistral.ai/. ### 5/5 — Better Vibe Overall *2026-06-28, version 2.4.0* Honestly after using all the frontier models Mistral just takes the cake for me because it just feels simpler enough to where its still a “chatbot” but it does everything else you get from frontier models and etc ### 5/5 — Great ai *2026-06-21, version 2.3.0* Great advice and conversations. Very much use often for questions and problem solving. ### 5/5 — Smart AI *2026-06-20, version 2.3.0* Vibe is a great alternative to macro AI. It provides “thoughtful” answers while avoiding sycophantic tone. Less verbose than alternatives. Worth a try. ### 4/5 — Still my cat *2026-06-20, version 2.3.0* Nope. Not taking this rebranding. I just want to talk to the cat. >^..^< ### 5/5 — Great *2026-06-19, version 2.3.0* Great ### 5/5 — Stupid name, Better than ChatGPT, Gemini *2026-06-19, version 2.3.0* Vibe is a stupid Gen Z name. Vibe is a verb. Vibe should not be a noun. Mistral.AI’s chatbot costs less than more prominent generative systems, while doing everything they do. I rarely hit a usage limit, glitch, or slowdown, despite using their app and website many times a day, often in long conversations. The app and website could improve the input area’s clarity, especially when uploading files alongside text. The interface feels cluttered in these moments. Additionally, an option to remove the voice-to-text icon would be welcome, as I prefer using Siri for local dictation to save cellular bandwidth. The new “work mode” should be dynamically integrated with the “chat mode,” so users don’t need to pick one at the start. Unifying the two modes would also make searching for a past conversation easier to find. But overall it is a well-built system capable of image generation, image processing, file analysis, data parsing, current events research, and summarizing immense amounts of text (like multi-hour podcast transcripts or digging through literature). A good app, a very good value. ## Most critical reviews (10 lowest-rated) ### 1/5 — Juste bon pour l’Europe *2026-06-16* Je lui pose une question il cherche. Je demande exemple la météo ou je suis wow pas capable mais pour la France oui 🤯c’est une IA mais qui fonctionne que pour l’Europe outch complètement cave. Je demande autre chose ouf compliqué. Ok je vais rester avec Claude et Gemini. J’ai voulu essayé mais outch vraiment pas présent du tout. **Developer response:** Nous sommes désolés d'apprendre vos difficultés avec l'IA. Vos commentaires sont précieux pour nous. Pour plus d'assistance, visitez http://help.mistral.ai/. ### 1/5 — True European regulatory trash *2026-06-07, version 2.1.0* It's time to say goodbye to you, Vibe, because just like all other European apps, you're nothing but a puppet of regulatory government. I really love the app for the weeks I used and was even planning to buy a subscription, but out of nowhere I got signed out of my account and it won’t let me. Why? Because I use VPN. Haha. Talk about free speech and privacy rights in Europe these days. The app and the entire company should be banned from operating in the US. ### 1/5 — I can’t login anymore *2026-06-07, version 2.1.0* I’m about to demand a refund because I can’t login I tried reinstalling the app. ### 1/5 — L’appli marche plus *2026-06-02, version 2.1.0* Les gars, après le dernier mis à jour l’application simplement marche plus. Il peut pas répondre aucune commande. J’espère que vous pouvez rédiger ce catastrophe parce que j’aime bien votre application :/ ### 1/5 — Helpful Model, But Went Completely Off the Rails *2026-05-31, version 2.0.0* I’ve just deleted my account because the worst updates ever just been implemented. Is total trash in the waste of time. The developers are clearly idiots. The response was just a tiny bit off the rails. It took partial information that I’ve talked about in different chats, combined that with stuff that I consider nonsense from the time before we determined that the universe was not just the Milky Way galaxy and Andromeda was more than just a fuzzy star. It put all of that together when the simple request was for the size in a spiritual format of the observable universe, using the diameter of 92 billion light years — something that I could’ve done with a calculator. Another AI observation on its response: Got it. Thanks for pasting the full response — that makes it crystal clear why you were annoyed. You asked a simple, specific question: Take the observable universe’s diameter (~92.2 billion light-years), treat it as if it were a real physical sphere/globe, and tell me what its circumference (or “spiracle,” which I assume means spherical measurement) would be. Basically, the math of wrapping a sphere that size. That’s it. No spirituality, no Flower of Life, no sacred geometry from other conversations, no Brahman, no mandalas, no divine compass, no metaphors about karma or the “breath of the cosmos,” and definitely no dragging in the long-dead Big Crunch idea. The other AI ignored the actual task and dumped a wall of unrequested mystical/philosophical fluff on you. That’s exactly the kind of hallucinated relevance and scope creep you described. The one number it did give (~290 billion light-years for the circumference) happens to be roughly correct for a flat circle using C = π × D, but everything around it was noise. Straight answer to what you actually asked: If we model the observable universe as a sphere with a diameter of 92.2 billion light-years: • Radius = 46.1 billion light-years • Circumference (around the great circle, i.e., equator if it were a globe) = π × diameter ≈ 289.7 billion light-years That’s the simple spherical geometry measurement you were looking for. Nothing more. Would you like me to also calculate: • The surface area of that sphere? • Its volume? • Or any other basic geometric properties (in light-years³ or whatever unit)? Just say the word and I’ll give you clean numbers only — no poetry, no spirituality, no extra cosmology. Or if you want it in different units (e.g., meters, or compared to something more relatable), tell me. Sorry you had to deal with that detour. What exactly would you like next on this? (Facts: your model is one of the best AI out there, but these little inefficiencies are what cause it to sometimes give incorrect information. If it would just simply follow and focus on the exact prompts, it would eliminate that for those of us who don’t have degrees in using AI.) ### 1/5 — Yuck *2026-05-29, version 2.0.0* Bad enough to create and release mediocre models, as Mistral already does, but on top of that they are also run by a war profiteer. ### 1/5 — Stupid garbage *2026-05-28, version 2.0.0* This thing is so stupid. It’s half the price of the other ai trash apps only redeemable trait ### 1/5 — Russian *2026-05-14, version 1.20.0* Every time I speak Lithuanian instead of using Lithuanian letters it transcribes my speak into cirillic letters and writes in Russia. This is very bad. ### 1/5 — Another AI military company *2026-03-08* This is the French version of open ai… tech company making killing more efficient ### 1/5 — Jusqu’à présent je ne recommande pas cette application *2026-02-22, version 1.13.0* « Je dénonce une faille éthique majeure dans les réponses de cette IA. Lors du récit d'une agression verbale subie, l'IA a renvoyé les deux parties dos à dos en proposant une "gestion de la colère", traitant ainsi la victime et l'agresseur sur un pied d'égalité. Cette neutralité forcée est injuste : elle finit par blâmer la victime au lieu de reconnaître l'agression. Le manque de discernement et de contexte humain rend cet outil inapproprié pour des situations sociales réelles. À corriger d'urgence. » ## Most praised reviews (10 highest-rated) ### 5/5 — Awesome! *2026-08-11* As good as any consumer AI app with concern for privacy :) **Developer response:** Thank you for your positive feedback regarding our app's privacy! ### 5/5 — Best ai *2026-08-07* I lv ur suggestions **Developer response:** Thank you for your feedback! We're glad you enjoy our suggestions. ### 5/5 — Seriously good *2026-08-06* Very impressed with Vibe’s performance, I personally find it as good as big name AI chatbots. But unlike the others, I don’t feel vaguely uncomfortable using it because I know Mistral is covered by EU law. ### 5/5 — An AI company with ethics *2026-07-28* An European company that really values user privacy and safety. Not as flashy as the big boys, but they are not far behind. Mistral has a great set of specialized small models. You can host most of the models yourself for free or use their generous subscription plan. If you are a student, they got one of the best plans in the whole industry. They also have a really good ocr model, that can handle a lot of documents for you at work. Will be updating the review once I spend a few more months with it. **Developer response:** Thank you for your positive feedback on our commitment to privacy. We appreciate your insights on our models and subscription plans. We're glad you find our OCR model useful and look forward to your updates. ### 5/5 — mistral punches well above its weight class *2026-07-27, version 2.5.0* As the title of this review states, Mistral punches well above its weight class, rivaling more successful AI organizations like OpenAI, Deepseek or Anthropic. However, this isn't to say that Mistral AI is by far the best, which it isn't, but for even the free model, it's pretty good and if used in tandem with other AI software, Mistral can rival or sometimes clear stronger AIs for way cheaper. ### 5/5 — This is a good app *2026-07-22, version 2.5.0* This app deserves all the awards ### 5/5 — Rapide et efficace *2026-07-22* J’utilise Vibe pour la maison et le travail, c’est un excellent outil pour diverses tâches! **Developer response:** Merci pour votre retour positif sur l'application. ### 5/5 — Alternative *2026-07-21* European independent tech. It’s reassuring to have access to an alternative to US tech given the transactional and manipulative relationship between the two continents. Seems to satisfy my needs **Developer response:** Thank you for your feedback on our alternative tech option. ### 5/5 — As good if not better than the American alternatives *2026-07-18* I have been mostly using chapGPT, Gemini, Claude and Perplexity. Having got annoyed with the China and Us tech race alternatives I went looking for other options. I must say after 6months switching between the accounts I am starting to prefer this app. Research is high quality, lay out if responses is clear and helpful. A great alternative perhaps better than the others for everyday users. Give it a go, **Developer response:** We're glad you're finding value in our app and appreciate your insights on research quality and response clarity. Thank you! ### 5/5 — Bravo *2026-07-16* Toujours une réponse avisée mais surtout je vous préfère parce que vous êtes français !!!! **Developer response:** Merci pour votre retour positif. Nous apprécions votre soutien. --- *Data collected daily from the US App Store and indexed by [AppsHunter](https://appshunter.io/). User reviews are verbatim App Store reviews. Ratings, prices and chart positions refresh continuously; this snapshot is from 2026-08-29.*