59 providers from one engine
Quota, spend, reset windows, token totals, request totals, and local-model server state share one Rust provider core.
v0.4.32 test build - 59 providers - desktop + mobile
See every AI provider’s limits, credits, and spend in one local-first dashboard.
LLM Status keeps rate limits, balances, spend, token totals, request totals, and local-model server state visible across 59 provider integrations on Windows, macOS, Android, and iOS.
Private test builds only. Read platform-specific install notes.
59providers
4platform tracks
0product telemetry
Why it is different
Built for people who bounce between models all day.
Several accounts, one view
Every account keeps its own credentials, history, and alert state, so two Claude, Codex, or Copilot plans never blur together.
Native widgets and refresh
Android home-screen widgets and iOS WidgetKit previews keep critical providers visible between app opens.
Desktop command center
Tray and menu-bar status, usage history, threshold alerts, a GitHub developer dashboard, and optional InfluxDB-compatible metrics export.
Pair a phone without a server
One device shows a six-digit code and the other finds it on your own network. Both prove they know the code, and every later transfer is encrypted end to end.
Local-first privacy model
Credentials stay in platform secure storage. There is no product telemetry, and this site runs no analytics or tracking cookies.
Platform coverage
One provider engine, native surfaces everywhere.
Windows and Android carry the current test packages. Every client shares the same Rust engine, so provider coverage does not depend on which screen you open.
Windows
Tray flyout, dashboard, usage history, alerts, backup and import, credential scan, device pairing, and optional metrics export.
macOS
Menu-bar app on the same Rust engine with Keychain storage and a template icon. Apple packaging is gated on Apple build capacity, with production trust gated on Developer ID signing and notarization.
Android
Native Compose test client with encrypted credentials, account slots, alerts, history sparklines, home-screen widget, WorkManager refresh, and PIN-protected config transfer.
iOS preview
SwiftUI Simulator preview with Keychain storage, multiple accounts, history sparklines, local alerts, WidgetKit widgets, secure OAuth refresh, and PIN-protected config transfer.
59 provider integrations, including local models
Reuse supported sessions and keys, connect with OAuth/device flows, paste API credentials, or monitor local Ollama, LM Studio, and llama.cpp servers.
- Anthropic Claude
- OpenAI Codex
- OpenAI Platform
- GitHub Copilot
- GitHub Actions
- Google Gemini
- Cursor
- OpenRouter
- DeepSeek
- ElevenLabs
- Deepgram
- Groq
- xAI Grok
- z.ai
- MiniMax
- Mistral
- Moonshot AI
- Fireworks AI
- StepFun
- Cerebras
- Rev.ai
- LMNT
- Aihubmix
- Alibaba Cloud
- Anthropic Console
- Cloudflare Workers AI
- Hugging Face
- Novita AI
- NVIDIA NIM
- SiliconFlow
- Vercel AI Gateway
- Xiaomi MiMo
- Poe
- You.com
- Portkey
- Requesty
- LangDB
- LiteLLM
- Bifrost
- CheaperInference
- Perplexity API
- Perplexity Search
- Brave Search
- SerpAPI
- ScrapingBee
- Apify
- Tavily
- Firecrawl
- Meshy
- Tripo3D
- Rodin (Hyper3D)
- Stability AI
- Higgsfield
- Nous Research
- Ollama
- LM Studio
- llama.cpp
- OpenCode
- OpenCodex
Privacy and trust
Your provider credentials stay under your control.
LLM Status uses secure operating-system storage, talks directly to the providers you configure, and does not collect product telemetry.
Secure local storage
DPAPI, Keychain, Android encrypted preferences, and iOS Keychain keep secrets off the marketing site.
Direct provider access
Credentials are used only to request usage information from the providers you connect, and they are not forwarded to a different origin.
Open implementation
Review the MIT-licensed source, architecture, provider fact sheets, and release gates in the repository once you have access. View source.
New in this test build
Pair your phone with a code on your own network.
Under Settings, one device shows a six-digit code and the other finds it on the local network. No account, relay, or cloud service sits between them.
Both devices know the code.
Each side proves it holds the six-digit code with SPAKE2, so a device sitting in the middle cannot join the pairing.
Nothing listens until you ask.
The listener starts only when you begin pairing, accepts local-network addresses only, stops after ten idle minutes, and five wrong codes end the attempt.
Config moves both ways, keys optional.
A paired device can send or fetch a configuration with or without API keys, and only after the other device's owner accepts the request.
Pairing is implemented source work in the desktop, Android, and iOS clients; testing between two physical devices is still pending.
Latest repository updates
The newest work hardens the data, not just the launch page.
Recent changes focus on encrypted transfer, archive ownership, truthful provider data, and credential handling.
PIN-protected configuration files
Exporting a configuration asks for a 4-6 digit PIN and encrypts the file with it (Argon2id plus ChaCha20-Poly1305). Desktop exports open on phones, and the reverse.
A usage archive you own
Usage history is kept in a local archive (the on-screen chart keeps up to 288 samples per account), exportable as credential-free JSONL and CSV, with a desktop JSONL restore preview.
Credentials stay where they belong
Provider credentials are no longer forwarded when a request redirects to a different origin, and the build carries current advisory fixes for its TLS and ZIP dependencies.
Truthful provider data
Providers without a verified read-only usage API report Unsupported instead of a fabricated zero, and fixture checks fail closed when a metric-bearing payload loses a required field.
Current release
v0.4.32 is a test build, not a store distribution.
The current build spans 59 providers, adds llama.cpp alongside Ollama and LM Studio, pairs nearby devices without a server, and encrypts exported configurations with a PIN. The signing and store gates below still stand between this build and a public distribution.
- Windows installers are unsigned, so SmartScreen may warn.
- Android packages are debug-signed until production signing is enabled, so upgrades may need a clean install.
- macOS packaging is gated on Apple build capacity, and this build ships no Apple package.
- iOS ships only as a Simulator package, and Apple build capacity is disabled for this release.
- Test builds are staged in the private repository, so installers require repository access.
Common questions
Before you install
Which platforms does LLM Status support?
The current test build ships Windows 10 1809+/11 and Android 8+ packages. A macOS 11+ menu-bar client and an iOS 16+ Simulator preview are built from the same source, with Apple packaging gated on Apple build capacity.
Where are credentials stored?
Credentials use secure operating-system storage and are sent only to the provider APIs you configure. LLM Status does not collect product telemetry.
Is this a subscription?
No. Test builds are shared privately with repository members. When paid early access is offered, it is a one-time $1.99 purchase through Stripe.
Are the current downloads production-ready?
No. Version 0.4.32 is a test build. Windows installers are unsigned, Android packages are debug-signed, and Apple packages are gated on Apple build capacity.
What is new in the current build?
llama.cpp joins Ollama and LM Studio as a local-server provider for 59 providers in total, nearby devices pair with a six-digit code and no server, exported configurations are PIN-encrypted, usage archives restore from JSONL, and desktop gains a GitHub developer dashboard.
How do I get a test build?
Test builds are staged in the private repository rather than published as public downloads, so installers are shared with repository members. Request access before following the platform install notes.
Current test build
v0.4.32
test build
59 providers, native desktop and mobile clients, device pairing, widgets, alerts, local servers, and MIT-licensed release notes.
View release notes and builds