Lumo alternatives: why people leave Proton's AI, and where the prompt goes instead

The Morse teamPublished October 29, 2026Facts checked October 29, 202610 min read

The best Lumo alternative depends on which of Lumo's limits sent you looking. As of October 29, 2026, Proton's Lumo runs only open models such as Qwen 3.5 and GLM 5.3, keeps a Free account's chats for 7 days, sells "unlimited" Plus under a fair-use rule, and has Proton's own servers read every prompt in the clear. Morse answers the model and the server problems in one account: GPT, Claude, and Gemini sent without your identity, and end-to-end encrypted chats your own browser checks. Duck.ai, Maple, Venice, Confer, Brave Leo, and a local app each answer at least one of them.

On October 29, 2026, we read Proton's Lumo page, pricing page, getting-started, privacy, export, and dictation help articles, the Lumo terms (last modified August 19, 2026), the Lumo privacy policy (last modified September 17, 2026), and the Lumo security model post, then each alternative's own pricing, privacy, and technical pages. We took each company at its word and ran no tests. Proton deserves credit first. It says plainly where its encryption stops, runs its models on servers it controls, and lets you export your chats on the way out. Our Lumo AI review covers Lumo itself in depth.

Lumo alternatives compared

As of October 29, 2026. Every cell for another app comes from its own page, linked in its section below; Morse's come from our privacy policy and plans.

Why do people look for a Lumo alternative?

Mostly for reasons Proton's own pages state.

No GPT, Claude, or Gemini. Proton's privacy help page lists the models behind Lumo as "Qwen 3.5, GLM 5.3, Image-Turbo, FireRed-Image-Edit-1.1, and Apertus 1.5," all open models run "exclusively on servers Proton controls" (Proton). If your work needs a closed frontier model, Lumo doesn't carry one.

A Free account's history lasts a week. Lumo's getting-started article lists the Free plan as "Limited messages & chat history (chats are saved for 7 days only)" (Proton). The pricing page shows the same plan without the 7 days, and its FAQ says Free has "usage limitations on chats, favorites, uploads, images, and projects" without putting a number on any of them (Proton).

"Unlimited" has a ceiling. As of October 29, 2026, Lumo Plus costs $12.99 a month, or $9.99 a month "Billed at $119.88 every 12 months" (Proton), and Proton's Lumo page sells it as the way "to get unlimited chats and the fastest responses" (Proton). The Lumo terms, last modified August 19, 2026, say "unlimited" means "no fixed numerical cap applies to ordinary, good-faith use," and that Proton may throttle or limit an account whose use "substantially exceeds the typical usage patterns of individual users of the same plan" (Proton).

Proton's server reads the prompt. Proton's security post says "we need to provide the LLM with cleartext messages," which is why it calls its design "user-to-Lumo (U2L) encryption" rather than end-to-end (Proton). The privacy page says prompts are encrypted "so only the Lumo GPU servers are able to decrypt them" (Proton). The pricing page's line that zero-access encryption means "only you can read your conversations" describes stored chats; the server that answers still decrypts each prompt. The review has the rest.

Some answers are filtered. The getting-started article says "Proton has implemented safeguards in Lumo to prevent harmful or offensive responses" (Proton).

Are the usual Lumo alternatives more private?

Not the ones the top-ranked lists put first. AlternativeTo's Lumo page, which ranks for "lumo alternatives," opens on October 29, 2026 with ChatGPT, DeepSeek, Mistral Vibe, Claude, Perplexity, and Google Gemini, and says nothing about where any of them decrypts a prompt or keeps your history (AlternativeTo). AlternativeTo doesn't mention that a Free account's chats last 7 days, and neither does Lumo's pricing page. The table below starts from the limit you're leaving instead.

1. Morse, for a server you can check and the models Lumo lacks

Lumo asks you to trust a Proton GPU server that reads your prompt in the clear and runs code no one outside Proton checks. Morse moves that check onto your device and adds the closed models Lumo doesn't carry.

• E2EE, for prompts you won't hand to a server you can't check. Before anything leaves your device, your browser checks the enclave's hardware attestation itself, comparing what's running with the versions pinned in the Morse app. Only then does it encrypt the message so that one enclave alone can read it. The enclaves are run by third-party providers, not by Morse, and our servers pass along ciphertext they can't read. The reply is decrypted only in your browser. Open the lock on any E2EE reply to see each check that passed, or download the full report. E2EE runs open-weight models such as GLM, Kimi, and DeepSeek and comes with Pro, at $19 a month. It rests on the enclave hardware and attestation working as designed, and it has no web search or image generation, so switch modes when you need either.
• Anonymous, for the GPT, Claude, and Gemini you couldn't get in Lumo. Our proxy strips your account, email, and IP address and sends the request under Morse's own account. The provider sees the question without knowing it's you, and may keep it for a limited time to check for abuse, under its own policy, so leave your name out of the prompt.
• Private, for open models like the ones Lumo runs. DeepSeek, Kimi, GLM, and xAI's Grok run on hosts that agreed to keep nothing, since Morse has zero data retention agreements with its inference, search, and other providers. Where a provider confirms zero retention on each reply, Morse checks for it and refuses a reply without it.

Whichever mode you pick, your history is stored only in your browser, encrypted, on Free as well as paid plans, not on Morse's servers. Opening it takes two keys: one that never leaves your device and one our server hands you at sign-in. Morse's servers don't keep or log your messages. We don't train AI on your content, sell your data, or show ads, and the site runs no advertising trackers or third-party analytics. Because we don't store chats, we usually have very little to give authorities. Free gives you 15 messages a day with GPT-6 Luna, Claude Haiku 5.5, GLM 5.3 Flash, and more, 3 images a day, and web search, with no card. Pro adds unlimited Pro models, unfiltered models, memory across chats, and E2EE; Max is $49 a month. Clearing your browser deletes your chats for good, and we can't recover them, so keep a copy of anything you need.

2. Duck.ai

Duck.ai is the closest swap if what you missed in Lumo was GPT and Claude. Its free models include Claude 4.5 Haiku, GPT-6 Luna, GPT-5.4 mini, and Mistral Small 4, with Claude Sonnet 5.5 on Plus and Claude Opus 4.8 on Pro (DuckDuckGo). DuckDuckGo removes your IP address so requests appear to come from it, but says "Model providers need to see your prompts in order for their models to provide responses" (DuckDuckGo). Providers agree to "Never use your Duck.ai conversations to train their models," and two free models, gpt-oss-120b and Gemma 4 31B, run in a TEE where "Tinfoil cannot see your prompts or the model's response" (DuckDuckGo). Chats are saved locally on your device by default (DuckDuckGo). Plus is $9.99 a month and Pro $19.99 in the US (DuckDuckGo), both bundled with a VPN, and DuckDuckGo says "we don't offer a way to only subscribe to one feature" (DuckDuckGo). More in Duck.ai alternatives.

3. Maple

Maple fixes Lumo's server problem on its Free plan, which lists "End-to-end encryption" and web search (Maple). Its proof page says chats are "decrypted only inside the enclave," and that the app "runs this check at the start of every session," comparing the enclave's fingerprint with a list built into the app. "If any check fails, no encrypted session is created and nothing is sent." Maple also names the limits. A match "means Maple approved that software, not that it was built from the public source," and in a browser the checking code "is served by Maple each time the page loads" (Maple). Pro is $20 a month for "All AI models including Kimi, GLM, DeepSeek, and GPT-OSS," and the models on its price list are all open ones, with no Claude, GPT-5.x, or Gemini (Maple). That list includes a DeepSeek V4.1 Flash Uncensored model. Chats are encrypted on your device, sent to Maple's servers, and "synced automatically between devices" (Maple), with apps for iPhone, Android, Mac, Windows, and Linux (Maple).

4. Venice

Venice gives you frontier models and enclaves in one account. Its public list has 141 text models, Claude, Gemini, and GPT among them (Venice), and your history stays on your device, where "Venice never stores it on its servers." For its Anonymous models it advises "you should assume the provider is storing your content," and "TEE and E2EE models require a Venice Pro subscription" (Venice). Free is 10 text prompts and 15 image prompts a day, and Pro is $18 a month (Venice). Its privacy policy, effective June 2, 2026, says "We use analytics services, such as Google Analytics," and names "analytics, advertising, and security partners" among those that may use cookies and pixels (Venice). Venice vs Maple sets the two side by side.

5. Confer

Confer encrypts every message. Its privacy policy, last updated October 2026, says "All messages you send are end-to-end encrypted before leaving your device," that Confer "cannot decrypt or otherwise access the content of your messages," and that history "is stored encrypted on our servers, but only you hold the keys" (Confer). Your client verifies the enclave's quote and "confirms they match a release in the transparency log" (Confer). The App Store listing is free with a $34.99 Confer Membership, for a period it doesn't state (App Store), and the pages we read don't name its models. Maple vs Confer compares their keys and checks.

6. Brave Leo

Leo needs "No account or signup" for its free tier, says it "doesn't retain or share chats, or use them for additional model training," and keeps history on your device (Brave). Since October 15, 2026, chats with GLM 5.3 Flash are end-to-end encrypted into NEAR AI TEEs, though Brave says "Today, your browser relies on the gateway for this check" (Brave). Premium is $14.99 a month or $149.99 a year, with a 7-day trial (Brave).

7. A local app

LM Studio runs open models on your own computer. Its privacy policy, effective June 2026, says that for local use "none of your messages, chat histories, and documents are ever transmitted from your system," and that it "can run entirely on-device and offline"; its separate cloud features process requests on its servers (LM Studio). Local models are on the $0 plan, and the cloud plans start at $20 a month (LM Studio). Our guide to local AI vs cloud AI covers the hardware trade.

Choose Lumo or Morse

• Choose Lumo if you already pay for Proton, want history synced across devices with zero-access encryption, want Projects that read a linked Proton Drive folder, or want every model run on servers Proton controls under Swiss law.
• Choose Morse if you want your own browser to check the enclave before a prompt leaves it; GPT, Claude, and Gemini without your name on them; history kept encrypted in your browser, on the Free plan too; and no ads or training. E2EE comes with Pro, at $19 a month.

Send a first message from the home page without an account, or start free with 15 messages a day and no card.

Morse publishes this comparison. We've tried to keep it fair; if something's wrong or out of date, email hello@trymorse.ai.

Proton and Lumo are trademarks of Proton AG. Morse is not affiliated with or endorsed by Proton AG.