Huge update to getbased just landed.

Use your favorite CLI agent as an AI provider.

That means you can use your existing subscription with Codex, Opencode (including free models), Grok, Claude (self-hosteding only for now), Hermes, and OpenClaw.

getbased now has its own tools these agents can call, on top of their own. That leads to much better output, including web search by default.

This feature requires a small companion locally installed that serves as a bridge between your browser and CLIs. And of course, it's open-source so you don't have to trust it, feel free to verify it.



The great thing about using your own personal agents, such as Hermes Agent or OpenClaw, in getbased is that you get your actual agent responding to you with its memory, its personality, and its knowledge about you.

Hermes Agent by NousResearch and OpenClaw are both available from localhost and remote gateway.

If your agent is using Honcho, for example, that setup is used in getbased as well.


Chat got some serious love.

You can now select models and set effort directly from the input area, drop pictures and files to import, and set context.

Conversations also show your profile picture and your agent’s, and the whole chatting experience is significantly better.

The sidebar now lets you create projects, manage them and move conversations with drag and drop.

Both the chat and the sidebar are flexible — resize them as you see fit.



Besides that, dashboard widgets now scale and fit automatically.

The whole codebase was audited with the new GPT-6 Astra and cross-reviewed with Opus 5.

Fable 5.1 available from all providers.

It gets 85% less rejects on biology related questions so might be worth to try it. Looks pretty strong here!

1st class integration for WHOOP is in.

Limited to Self-hosting for the time being as it needs active subscription for the hosted version.

v1.18.0 — Meals & Nutrition arrives

Track your meals, macros, nutrition and hydration with a vision model of your choice. Simply add a picture and run analysis.

You can also use product labels or add everything manually.

The precision of the analysis depends on the picture(s) itself and used model but Gemini 3.7 Flash and the new GLM 5.3 Flash both feels great and pretty cheap.



See trends of your choice from 7D to 1Y.

As a bonus, getbased will automatically compute your eating and fasting windows, feeding everything into your Light & Circadian and Diet & Digestion context cards.

You'll also see your Carbs & Fats fuel ratio as a proxy to estimate the infamous Randle's Cycle, the natural rate limiter of the cells.



Dashboard widget shows you the most important last 7 day values and is fully customizable. Add or remove values in the widget as you see fit.



If you want to geek out, there's a benchmark feature behind Debug mode.

Compare up to four models against your preferred or manual values reference.

Let us know which model you find the best for the use case.



Multiple bugfixes & improvements landed including first class support for @UnslothAI so you can get the most out of local inference.

Go try it and leave feedback here or on Github!

👉 https://getbased.health

Grok 4.6 available via all providers, go try it!

1.15.0 — A redesigned and expanded Genome is out!

- Genome is clearer and easier to explore. Findings now use simple risk, protective, trait, and reference associations, with deeper interpretation available only when you want it.

- More useful variants are recognized. The curated SNP catalog has been expanded with widely studied wellness and health associations, while genotype matching and DNA-file imports are more reliable.

- Evidence is easier to check. SNP findings include relevant publication links and clearer evidence context, plus a direct way to suggest a correction.

- Mitochondrial DNA has been improved. More maternal haplogroup subclades are supported, and the mitochondrial climate and environmental-fit view is more complete and easier to use.

- Ask AI stays focused without limiting the model. Routine context remains compact, while asking about a finding adds its relevant details and still lets your selected model use broader knowledge.

- Imports and related insights are safer. Re-importing DNA preserves manually added findings and mitochondrial data, with additional fixes for strand handling, vitamin D genetics, and light-related guidance.

v1.14.0 — Supplements and medications, rebuilt

- Keep active routines clean. Ended supplements and medications move to history and can be restarted anytime.

-Add products faster. Choose common units, import from links and label photos, and review the result before saving.

-See clearer product context. Active ingredients, capsule materials, excipients, fillers and laboratory results stay organized separately and reach AI only when useful.

- Review better mitochondrial evidence. Verified primary studies are grouped by compound with clear scope and limitations.

- Pause Sync without disconnecting. Pausing this browser keeps its identity and local Sync history. Edits made while paused stay queued and publish before incoming changes when you resume; disconnect/reset remains available under Advanced.

- Restore and rejoin safely. Profiles restored from a full backup are published before older relay deletions can apply, so recovered supplement and medication history does not disappear during the first Sync.

- Join Sync reliably on the hosted app. Production builds now load the encrypted Sync database worker from its stable app location instead of waiting on a missing asset.

- Large Local AI lab imports finish more reliably. Progress now separates reading the report from writing results, long model-prefill waits no longer look like broken streams, and incomplete output is rejected instead of offering a partial marker list.

- Bug fixes and improvements. Product-page extraction and quality-result edits are more reliable, mobile layouts work better, and existing data migrates safely.

Chat improvements are in!

Go try it at https://getbased.health 👈

A completely redesigned health context cards just landed

- Health context has been redesigned from the ground up. The cards are cleaner, easier to scan and edit, work better on mobile, and explain why each area matters.

- The questions are more useful and more complete. Medical and family history, diet and digestion, exercise and physiotherapy, sleep, stress, relationships, and environmental exposures now capture the details most likely to affect interpretation.

-You stay in control of how much you add. Start with the essentials, add deeper context only when it is relevant, or use a custom answer when the available choices do not fit.

- Context is easier to interpret over time. Health priorities are clearer, important changes are preserved, and AI receives a more useful picture of what may be shaping labs and health patterns.

- Both demo profiles have been rebuilt. Alex and Sarah now include realistic, up-to-date context, labs, trends, wearables, light data, and other details that make the full app easier to explore.

- Demo AI is safer and more flexible. Ready-made demo insights use no AI. Local AI can update edited cards automatically, while paid AI is only used after clear permission.

A lots of work in the background indeed.

Hopefully finished soon and on to new features.

When was the last time you saw a health app with decentralized, E2EE open-source AI paid for with cypherpunk cash?😎

Verifiable end-to-end encrypted inference with @PayPerQ just landed in getbased.

So yes, you can now try GLM 5.2 privately.

No email.
No KYC.
No fiat.
No subscription.

Just pay per query and keep your prompts encrypted.