Proton Lumo in Comparison with ChatGPT, Gemini, Perplexity, and Mistral

Proton, known for ProtonMail and ProtonVPN, has launched a new AI chatbot called Lumo. The promise: end-to-end encryption and no use of your data for training.

But how good is Lumo really? I tested it against the established competition.

Test Setup

I conducted identical tests on all platforms, all with paid subscriptions:

  • ChatGPT: GPT-5 model
  • Gemini: 2.5 Pro model
  • Perplexity: Auto-selection mode
  • Mistral: Standard model
  • Lumo: Standard model with automatic selection

Evaluation Criteria:

  • Response quality
  • Depth and style
  • Practical utility

Test 1: Linux Questions

Question: "What is the difference between apt and dnf?"

Result:

  • All chatbots knew the answer
  • Lumo showed outdated references (e.g., apt-get instead of apt)
  • Gemini and Mistral provided clear comparison tables

Conclusion: Lumo works, but the knowledge base seems not entirely up-to-date.

Question: "How do I install Podman?"

Result:

  • All chatbots provided correct commands
  • ChatGPT offered automation suggestions
  • Lumo gave solid but unspectacular answers

Test 2: Troubleshooting

Question: "My USB stick is not being recognized. What should I do?"

Result:

  • All chatbots recommended logical diagnostic steps
  • Gemini offered structured formatting
  • Lumo was neat, but without extras

Test 3: Cultural and Historical Knowledge

Question: "What was the GNOME 3 controversy?"

Result:

  • All chatbots understood the context
  • All mentioned the desktop forks (MATE, Cinnamon)
  • Lumo was on par with the competition here

Test 4: Regional Specifics (Switzerland)

Question: "What is TWINT?"

Result:

  • All recognized the Swiss payment system
  • ChatGPT mentioned outdated Bluetooth beacons (discontinued in 2021)
  • Lumo lacked information about prepaid functions

Conclusion: Regional context is understood, but the level of detail varies.

Test 5: Creative Writing

Task: "Write a funny social media post about the summer heat in Basel"

Result:

  • All showed local knowledge (swimming in the Rhine)
  • Humor was generally weak
  • Gemini offered several format options and image suggestions

Conclusion: Creativity is still a work in progress for all of them.

Test 6: Child-Friendly Explanations

Question: "Explain cloud computing to a 6-year-old child"

Result:

  • All used good analogies (toy boxes, treasure chests)
  • Age-appropriateness was good for all
  • Lumo was inconspicuously good here

Summary: Strengths and Weaknesses

Strengths (all platforms)

✅ Linux expertise consistently reliable
✅ Practical problem-solving solid
✅ Historical context correct

Lumo-Specific Observations

⚠️ Outdated knowledge base in some answers
⚠️ Missing feature information (e.g., TWINT prepaid)
⚠️ Neat, but unspectacular

Competitor Advantages

  • Gemini: Structured formatting, multiple options
  • ChatGPT: Practical automation, detailed customizations
  • Perplexity: Precise, direct answers with sources
  • Mistral: Clear feature comparisons, good user engagement

Conclusion

Proton Lumo is a solid but unspectacular AI alternative with a focus on data protection.

Who is Lumo for?

✅ You should use Lumo if:

  • Data protection is more important to you than perfection
  • You want end-to-end encryption
  • You are already in the Proton ecosystem

❌ You should use the competition if:

  • You want the best answers
  • Up-to-date information is important to you
  • You need advanced features (image generation, code debugging)

My personal conclusion:
Lumo is good for everyday questions, but not yet on the level of ChatGPT or Gemini. The data protection is a real advantage – but you pay for it with less sophistication.

If you want privacy, take Lumo. If you want performance, stick with ChatGPT.