Almost there
The 2.0 update has been huge and it now feels possible to jump from frontier model providers to an alternative that comes close in performance and features whilst keeping your data private.
There’s still a few niggling issues (the load time when first opening the app on iOS is way too long, not being able to add existing conversions to projects, not being able to have project-specific personalisation, custom lumos not being available on the app). Lumo will be great if you address this sort of things and keep up with the latest/best open weight models.
There’s still a few niggling issues (the load time when first opening the app on iOS is way too long, not being able to add existing conversions to projects, not being able to have project-specific personalisation, custom lumos not being available on the app). Lumo will be great if you address this sort of things and keep up with the latest/best open weight models.
Show more
Unusable
Tap mic button – immediate crash, app closes. Not occasionally, every time. If you cannot talk to your AI, it is not useful. It’s 2026 – why do we even have a text field anymore? (I’ll give it to you for odd spellings and exact numbers, but honestly … )
Lumps 1.0
I downloaded Lumos 1.0 when it came out and liked it just the way it was. Proton upgraded my app. I would like to have Lumos 1.0 back on my device. Is that possible?
Really good but could use refinement
I value privacy so I downloaded the app. It works good for my daily usage . The responses aren’t as verbose as chat gpt. I would like the ability to upload multiple pictures at once. Also I think it would be good to not have to enter a prompt when a picture is uploaded within the same chat.
Depends on use case
It certainly chats with you. I wouldn’t use it for research. I mainly use ChatGPT. I asked ChatGPT to come up with a knowledge test for Lumo, and it generated a prompt to determine who oversees the CDC and the WHO and what actions have been taken by either in the past 90 days. Lumo took a really long time to think about it before giving up (based on other reviews, I’m not sure if giving up was an error or because the prompt was political, though I wasn’t trying to make the AI take any particular stance).
Then I decided alright, let’s ask it something simple but somewhat obscure, so I asked it for Sonic’s middle name. It said Sonic did not have a middle name, not even in the Archie comics. ChatGPT got Maurice immediately and provided a citation.
I’d be comfortable using this to bounce ideas off of or vent, but it’s ability to search for information isn’t quite up to par. I do really like the interface and the cat mascot though. I hope it continues to improve.
Then I decided alright, let’s ask it something simple but somewhat obscure, so I asked it for Sonic’s middle name. It said Sonic did not have a middle name, not even in the Archie comics. ChatGPT got Maurice immediately and provided a citation.
I’d be comfortable using this to bounce ideas off of or vent, but it’s ability to search for information isn’t quite up to par. I do really like the interface and the cat mascot though. I hope it continues to improve.
Show more
Great AI assistant, but strong bias towards Proton products
I love the privacy-focused aspects of Lumos and it works well. The biggest drawback to me is that the model is strongly biased towards recommending Proton products, particularly when asking for product comparisons. This raises a significant ethical concern about how this model is trained. I would like to see more transparency about model training as well as updated configuration to prevent biased responses that benefit the parent company.
Show more
Keep the cat 💜🫶🏻
I saw someone saying they hated the cat and called it creepy. No sybau I love Lumo pls keep the cat. 🙂↕️💕
I’m still using the free plan and I’m primarily using it for venting with light researching or some brainstorming and chats. Just keep in mind that it’s still outdated (2024 I think) which is disappointing so you’ll have to double check info on your own or ask it.
I’m not sure what the character limit is for personalization or memory (I forgot what it was called) but I think it’s definitely seems more than ChatGPT because I tried to copy & paste then added a lot more instructions and so far I think it remembers more than gpt (also on free plan)
I’m not gonna sugarcoat it though I don’t fw the image generating Maybe I didn’t experiment enough. I can’t really comment on other things since I mainly just vent.
Overall I’m enjoying it but I’m hoping for more improvements and new features added. But pls creating images needs work imo.
I’m still using the free plan and I’m primarily using it for venting with light researching or some brainstorming and chats. Just keep in mind that it’s still outdated (2024 I think) which is disappointing so you’ll have to double check info on your own or ask it.
I’m not sure what the character limit is for personalization or memory (I forgot what it was called) but I think it’s definitely seems more than ChatGPT because I tried to copy & paste then added a lot more instructions and so far I think it remembers more than gpt (also on free plan)
I’m not gonna sugarcoat it though I don’t fw the image generating Maybe I didn’t experiment enough. I can’t really comment on other things since I mainly just vent.
Overall I’m enjoying it but I’m hoping for more improvements and new features added. But pls creating images needs work imo.
Show more
Good Model
I like this model and have started using it as my main in personal life. Would love to see a functional widget to quick launch dictation, camera, or text chat. Would also love to be able to trigger dictation with the action button. Only pain points vs claude or chat.
Still Needs a lot of improvement
I’m sorry but this AI model keeps getting stuff wrong. Even when asking about things that are easily available on the internet. I have plenty of examples of it making stuff up and then back tracking once I call it out. Also Proton keeping their cloud storage subscription separate isn’t a convenient. Hopefully they make some much needed changes and I can try this again in the future.
Show more
Constantly upgrading while keeping your privacy.
I used chat GPT and Gemini before switching to Lumo. The collection of data those companies use mixed with complete lack of privacy, while selling said data to the highest bidder is concerning to say the least. This bot is constantly upgrading and getting faster, it lags a little but I’ll take that over someone selling my data and tracking me. If it didn’t have the privacy feature I would only give 4 stars but this company is grinding to make it work and I appreciate them. More should give this a shot.
Show more
New Update Is Amazing
I am very satisfied with the latest update and I can’t wait to see what comes in the future and the best part is the privacy.
WTH
Won’t let me download Lumos saying I have to enter my password in order to download it. I already have a subscription for proton. Why won’t it let me download? this is bs
Getting better
Finally in the same intelligence ballpark as other tools using frontier models. Web search capability could be better, and the weird way most of the ui doesn’t load straight away.
Response from developer
Thank you for editing your review. We'd love to earn a 5-star rating from you - if you have any suggestions on how we can improve our service after you've tried it out, don't hesitate to pass your feedback along to us. We always appreciate our community's input.
lumo 2.0 shows no 7 day chat history
even with settings set to "show chat history" in free version (that promises to keep 7 days of chats), no history is shown under "today" in history tab. the conversations disappear quickly and app appears to be frozen. also max version does not turn on and gives no explanation. very disappointed with update.
Show more
Expérience utilisateur à peaufiner
Bonjour,
Dans les versions précédentes, le routage était géré automatiquement selon la complexité de la requête.
Avec la version 2.0, la sélection manuelle du modèle et du mode de réflexion déplace cette responsabilité vers l'utilisateur.
Le choix entre Lite et Max suppose une compréhension technique non je ne dispose pas et me semble un élément de complexité inutile.
Suggestion proposée : une approche hybride
Comportement par défaut : Routage automatique intelligent du modèle et du mode de réflexion selon la complexité détectée de la requête (comme c'était le cas antérieurement).
Réglages avancés optionnels : Un accès aux paramètres avancés permettant aux utilisateurs souhaitant un contrôle explicite de désactiver le routage automatique et de choisir manuellement leur modèle et mode.
Merci pour votre travail
Dans les versions précédentes, le routage était géré automatiquement selon la complexité de la requête.
Avec la version 2.0, la sélection manuelle du modèle et du mode de réflexion déplace cette responsabilité vers l'utilisateur.
Le choix entre Lite et Max suppose une compréhension technique non je ne dispose pas et me semble un élément de complexité inutile.
Suggestion proposée : une approche hybride
Comportement par défaut : Routage automatique intelligent du modèle et du mode de réflexion selon la complexité détectée de la requête (comme c'était le cas antérieurement).
Réglages avancés optionnels : Un accès aux paramètres avancés permettant aux utilisateurs souhaitant un contrôle explicite de désactiver le routage automatique et de choisir manuellement leur modèle et mode.
Merci pour votre travail
Show more
Response from developer
Thank you for your review. We’ve passed on your feedback to the team for further investigation. Meanwhile, if possible, you can try switching to a different network to see if the behavior improves.
Update not as good as before
1)The recent update changed the style of responses they are not as good. Whether using the max 2.0 or lite,how is this an upgrade? 2)The chat can no longer hold certain files as before ,the past month has had a significant memory decline, it keeps saying the chat has gotten to long and context length exceeded and shows token usage. The file size has to be very small like 200-300 MB,before the update there was no errors. It has changed twice over the last month.My group is using Proton Lumo for our school project but had to switch back to GOOGLE 1.5 Pro it holds 2,000,000 tokens content.
Show more
High Moderation
I am excited for the future of Lumo, but at the moment it is not ready. There are other private AI projects that work better currently.
The high moderation of Lumo also makes searching topics like parkour jumping can be dangerous and will not show results.
The high moderation of Lumo also makes searching topics like parkour jumping can be dangerous and will not show results.
Network issue !?
Unfortunately, the programme was working, but a network issue arose after the latest update; the fault lies with the company’s servers. Please resolve the issue
Poor performing AI
Tried this because of the new 2.0 version to see how this AI would perform but was disappointed. You claim privacy but no information is disclosed as to the AI itself as to how it’s routed, the LLM used or any training data, documentation and so forth.
2.0 is fantastic
The 2.0 version released in June 2026 is fantastic. On par with the best models for conversational work, plus much better privacy.
