Skip to main content
PromptQuorumBuilt for humans. Structured for AI.
Home/Power Local LLM/AnyLLM Review: BYOK AI Chat Client for Android
Mobile & Edge LLMs

AnyLLM Review: BYOK AI Chat Client for Android

·7 min read·By Hans Kuepper · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI
Read in:
🇺🇸en🇩🇪de🇫🇷fr🇯🇵ja🇨🇳zh🇪🇸es🇧🇷pt🇸🇦ar🇰🇷ko

AnyLLM is a freemium, closed-source Android chat client from Dotwel Labs that connects to AI providers with your own API keys — OpenAI, Anthropic, Google, Groq and others — and to an Ollama server for local models. It runs no model itself. Per its Google Play listing it needs Android 7.0 or later; this review covers version 1.5.1 and is based on the public listing, the developer's site and its privacy policy, not on hands-on testing.

AnyLLM is an Android chat app where you bring your own API keys for providers such as OpenAI, Anthropic, Google and Groq, and can also connect an Ollama server for local models. This review covers what its Google Play listing, the developer's site and its privacy policy document, what stays unclear, and how it compares with other mobile chat clients.

AnyLLM Review: BYOK AI Chat Client for Android

Key Takeaways

  • Maker: Dotwel Labs, an independent studio whose site presents AnyLLM as its flagship app.
  • Audience: people who already hold API keys and want several providers in one Android app.
  • Standout features: AI-vs-AI debates, MCP servers, SKILL.md agent skills and web search.
  • Limits: closed source, no on-device inference, and the listing does not say what the in-app purchases unlock.
  • Maturity: Google Play shows 1,000+ downloads, so it is a young app with little independent coverage.

📍 In One Sentence

AnyLLM, by Dotwel Labs, is a free-to-download Android chat client where you bring your own API keys for OpenAI, Anthropic, Google, Groq and others, with optional Ollama, MCP tool and web-search support.

💬 In Plain Terms

Think of it as one chat app for several AI services: you paste in your own keys, pick a model and pay each provider directly. It does not run AI models on the phone; the "local" part is an optional link to an Ollama server you run yourself.

Get AnyLLM

AnyLLM is distributed through Google Play. It is a free download with in-app purchases.

This review is a companion to PromptQuorum's Local LLM Software Directory, which lists AnyLLM next to other mobile chat clients.

Channel
Get It
Google PlayAnyLLM: AI Chat Client on Google Play
Websitedotwellabs.web.app
Privacy policyAnyLLM privacy policy

No public source repository was found, and there is no iPhone or desktop build; the listing covers Android only. The website is the developer's studio page, not a dedicated AnyLLM site.

AnyLLM at a Glance

Attribute
AnyLLM
PlatformAndroid 7.0 and later
Version1.5.1
Last app updateSeptember 5, 2026
PriceFree, in-app purchases
LicenseClosed source
Runs models itselfNo model engine is described; it connects to providers and Ollama
Image inputYes, images can be attached per the privacy policy
Voice input / outputNot stated in the listing
MCP tool serversYes, SSE and Streamable HTTP

Attributes follow the mobile-chat comparison used in the Local LLM Software Directory. "Not stated" means the Google Play listing and the developer's site do not say, not that the feature is absent.

What AnyLLM Is

AnyLLM is an Android chat client, not a model runner. The Google Play listing pitches "one unified workspace to talk to GPT-4o, Claude, Gemini, Llama, Groq, and more" and describes a bring-your-own-key (BYOK) model: you enter your own API keys and pay the providers directly.

It is published as Dotwel Labs, which the developer site describes as an independent product studio. Do not confuse it with AnythingLLM, an unrelated open-source app.

This review draws on the Google Play listing, the developer site and the privacy policy only. It has no hands-on testing, so speed, stability and answer quality are not rated here. Marketing lines such as "the most polished AI chat UI on Android" are the developer's own and are not repeated as findings.

How to Get Started

Setup is a Google Play install, then adding at least one provider key. The listing does not document the first-run flow, so the steps below follow what the listing and privacy policy state.

  1. 1
    Check your phone
    Why it matters: The listing requires Android 7.0 or later. No RAM or chip requirement is stated, because the models run at a provider or on a server rather than on the phone.
  2. 2
    Get your keys ready
    Why it matters: AnyLLM is BYOK, so create an API key at each provider you want to use. The privacy policy says keys are kept on the device in secure storage.
  3. 3
    Install and add a provider
    Why it matters: Install the app from [Google Play](https://play.google.com/store/apps/details?id=com.dotwellabs.anyllm), enter a key or a custom OpenAI-compatible endpoint, and choose a model.
  4. 4
    Optionally add Ollama, MCP servers and search
    Why it matters: Point the app at an Ollama server for local models, add MCP servers, and enable DuckDuckGo, Brave Search or SerpAPI for web search.

Features

AnyLLM's selling point is the set of tools around the chat. Everything below comes from the Google Play listing and the privacy policy.

  • AI-vs-AI debates. Two models argue a topic against each other in real time. The listing calls this a feature no other AI chat app offers, a claim this review did not check.
  • MCP tools. Connects to Model Context Protocol servers over SSE and Streamable HTTP, with an agent loop that supports tool calling.
  • Agent skills. Import SKILL.md files to give a chat domain knowledge, for example a coding assistant or a writing partner.
  • Web search. Live browsing through DuckDuckGo, Brave Search or SerpAPI.
  • Reasoning mode. A thinking-mode toggle for models such as Claude and OpenAI's o1 shows step-by-step reasoning before the final answer.
  • Chat interface. Markdown, LaTeX math, code highlighting, streamed output and system-prompt presets such as Code Expert or Socratic Teacher.
  • Per-model controls. Temperature, maximum tokens and context-window size can be set for each model.
  • Memory. The privacy policy says the app extracts key facts from chats ("memory facts") and stores them on the device.

Providers, Ollama and Data Flow

Most of AnyLLM is a client for cloud APIs. The listing names OpenAI, Anthropic, Google, Groq, Mistral, DeepSeek and Together AI, and the privacy policy adds OpenRouter. Meta's Llama 3 is offered through Groq rather than run in the app. Custom endpoints and OpenAI-compatible APIs are listed, and per the listing you pay providers at their own rates with no subscription from AnyLLM.

Ollama is the local path. The listing describes an Ollama integration for "offline/local AI models", and the privacy policy says Ollama data stays on your local network or follows your Ollama cloud configuration. The listing does not say where the Ollama server must run; Ollama is a server program, so expect to run it on a computer you control and point the phone at it.

What goes where. Each message, and any image or document you attach, goes straight to the provider you selected, under your own agreement with that provider. With Ollama it goes to the server you configure.

Pricing and Privacy

AnyLLM is a free download with in-app purchases; the US Google Play listing shows $0.99 to $14.99 per item. Neither the listing nor the developer's site says what the purchases unlock, so check the in-app store before relying on a feature. Budget separately for API usage, which each provider bills.

The privacy policy, last updated April 12, 2026, says chat history, memory facts, API keys and model settings stay on your device in a local database or secure vault and are not uploaded to the developer's servers. It also says there is no cloud sync.

The policy discloses Google Analytics for Firebase: app launches, screens viewed, feature use, device model, OS version and country- or city-level location, but not chat content or keys. Opting out is possible in the system settings or by contacting the developer.

The app is closed source and this review did not inspect network traffic, so these statements are the developer's own.

Trade-Offs: Benefits vs. Limitations

One app, many providers

What it means in real use:
Switch between OpenAI, Anthropic, Google and Groq models in one chat interface.
Limitation / caveat:
Needs an API key per provider, and each bills separately.

Ollama connection

What it means in real use:
Use local or self-hosted models from the phone.
Limitation / caveat:
Needs an Ollama server you run; setup is not documented in the listing.

MCP tools and skills

What it means in real use:
Tool calling, SKILL.md skills and web search in a mobile chat.
Limitation / caveat:
Results depend on the model and the servers you add; not tested here.

AI-vs-AI debates

What it means in real use:
Compare how two models reason about one topic.
Limitation / caveat:
Each debate calls two models, so it uses more API credit.

Local chat storage

What it means in real use:
History and keys stay on the device.
Limitation / caveat:
Closed source, with analytics enabled; the claims are the developer's own.

AnyLLM vs. Alternatives

AnyLLM

Platforms:
Android
Price and license:
Freemium, closed source
Where models run:
Provider APIs and Ollama
Key difference:
Multi-provider BYOK with MCP, skills and AI-vs-AI debates

RikkaHub

Platforms:
Android
Price and license:
Free, MIT
Where models run:
Providers you configure
Key difference:
Open source, same provider-switching idea
Articles about RikkaHub (2)

Also mentioned in:

Solair AI

Platforms:
iOS
Price and license:
Free, closed source
Where models run:
On the device, optional own-key cloud fallback
Key difference:
iPhone only, local first
Articles about Solair AI (2)

Also mentioned in:

PocketPal AI

Platforms:
iOS, Android
Price and license:
Free, MIT
Where models run:
On the phone
Key difference:
Runs models on the device, no keys needed
Articles about PocketPal AI (10)

+24 more not shown

Platform, price, and feature details for third-party apps change frequently. Verify current specifics on each app's own listing before deciding.

Compare in directory

Who Should Use AnyLLM

  • Android users who already pay for API access. One app covers OpenAI, Anthropic, Google and Groq keys.
  • Tinkerers who want tools on the phone. MCP servers, SKILL.md skills and web search turn a chat into an agent loop.
  • Self-hosters with an Ollama server. The app can reach it from the phone.

Who Should Not Use AnyLLM

  • Anyone who wants models running on the phone. AnyLLM runs none; PocketPal AI and Private Mind do.
  • Anyone who needs open source. No repository was found; RikkaHub is MIT-licensed.
  • iPhone users. Only Android is listed; Solair AI is an iOS option.
  • People without API keys. Every cloud model needs a key and a provider account.

Frequently Asked Questions

Who makes AnyLLM?

Dotwel Labs, an independent studio that presents AnyLLM as its flagship app. The Google Play listing and the privacy policy are the only other public documents found.

Can AnyLLM run models on my phone?

No model engine is described. It is a client: models run at the provider or on an Ollama server. For on-device inference, see PocketPal AI.

Does AnyLLM work on iPhone or desktop?

No. The listing covers Android only.

Which search engines does web search use?

DuckDuckGo, Brave Search or SerpAPI, per the listing. The privacy policy names Brave Search as a third-party service.

Can I use my own endpoint?

Yes. The listing and the developer's site say custom endpoints and OpenAI-compatible APIs are supported, so a self-hosted gateway can work. Compatibility with a specific server was not tested.

Verdict

AnyLLM is a reasonable pick for an Android user who already holds API keys and wants OpenAI, Anthropic, Google and Groq models, MCP tools and web search in one app.

The caveats are age and transparency: it is young and small, closed source, and the listing documents neither the paid items nor the Ollama setup. It runs no model on the phone.

If you want models on the device, start with PocketPal AI. If you want an open-source multi-provider Android client, look at RikkaHub first, and try AnyLLM's free tier before buying anything.

Sources

← Back to Power Local LLM

MCP