NVIDIACanary unavailable
GLM-5.2
z-ai/glm-5.2Long-horizon coding and reasoning candidate in NVIDIA's catalog.
- Access
- Developer API access
- Protocol
- OpenAI compatible
Verified free APIs + a portable key workspaceFind a free API that works now.
Find a free API that works now.
Keep it ready everywhere.
Apply through the official path, verify the route, save your key once, use it across tools and get notified only when availability or free-access rules meaningfully change.
6 reviewed providers25 model routes2 live AI routes0 plaintext keys on server
Find an API key I can getCurated launch set
Useful models, with the access path attached
GeminiCanary rate-limited
Gemini Flash
gemini-flash-latestGeneral-purpose multimodal model through Google AI Studio.
- Access
- Account-specific free tier
- Protocol
- Gemini native
GroqCanary healthy
Qwen3.6 27B
qwen/qwen3.6-27bFast coding and reasoning model with documented Groq limits.
- Access
- Free-plan limits
- Protocol
- OpenAI compatible
OpenRouterCanary healthy
Free Models Router
openrouter/freeAutomatic routing across currently available free models.
- Access
- Completely free router
- Protocol
- OpenAI compatible
Follow the change, not every check.
Choose a provider or exact model, then receive recovery, sustained failure and access-rule updates on your schedule.
Build a Watchlist →Portable VaultSave once. Import and export safely.
Bring in .env, JSON, TOML-like configs or encrypted backups, then export keyless profiles for your tools.
Open Import & export →Direct PlaygroundLoad a saved key without another setup lesson.
Select a model, unlock the matching Vault item in place and send a real browser-direct request.
Try a model →One continuous workflow
From “I heard it is free” to a working setup
No dead-end directories. Each step carries the provider, model and evidence into the next one.
Check eligibility
Card, phone, region and account requirements.
Get your own key
Follow the current official application flow.
Verify safely
Test authentication directly against the provider.
Compare real output
Run the same prompt and measure actual latency.
Install the config
Generate the right format for your AI tool.
Cross-device by design
Your setup should follow you, not stay trapped in one browser.
Browse anonymously. Sign in when you want encrypted key sync, saved configurations, test history, alerts and device control. Signing in is an upgrade to continuity, never a gate to discovery.
Synced workspaceZero-knowledge vault
DesktopCodex + OpenCode configsActive
Browser6 reviewed provider profilesSynced
MobileSignup progressReady
Blog & verified updates
Teach the task. Publish changes only when they matter.
Practical guides explain the setup; dated updates cover free models, quota changes, key-flow changes and incidents.
Browse all articles →