Free AI API Distribution
Bring Your Own Key
Consolidate your OpenAI, Anthropic, Gemini, and DeepSeek endpoints into a single, secure gateway. Limit spending by model, audit query history, and test payloads inside our live tester. Completely free.
Server-Side Rest Encryption
Keys are encrypted using bank-grade AES-256-CBC and decrypted only at the server level during active API routing. We never expose tokens to client-side code.
Granular Spending Thresholds
Stay in complete control. Configure specific hard-spent limits on a per-API-key and per-model basis to safeguard your budgets from run-away loops.
Unified Request Auditing
Audit prompt and completion tokens, latency, status codes, and financial costs globally across all providers. Filter, search, and view logs immediately.
Integrate in Three Simple Steps
Start distributing requests with full safety and metrics in under two minutes.
Secure Your API Key
Log in and add your API key (e.g. OpenAI or Gemini) to the dashboard. The system tests connection immediately.
Configure Cost Limits
Select models, toggle activation rules, and define a budget limit (e.g. max $5.00 limit for GPT-4o models).
Direct Proxied Requests
Point your scripts to our secure endpoint /api/proxy. We append your key, run validation, and forward the results.
Supported AI Gateway Providers
Connect dynamically to the world's most powerful language models and specialized code autocomplete assistants.
Frequently Asked Questions
Yes, absolutely. DevDock is completely free to use. Because you bring your own AI API keys (BYOK), you only pay the respective AI provider (OpenAI, Anthropic, Gemini, etc.) directly according to your actual developer token consumption. We add zero platform markup or platform surcharge.
DevDock routes your request payloads directly to providers completely free. By using your own keys, you avoid paying SaaS markups, retain full ownership of your developer accounts, and pay only raw API costs directly to the respective providers.
Security is our core priority. All API keys are immediately encrypted at rest using server-side AES-256-CBC with a hardware-secured environment variable key. The raw keys are never sent to the browser and are only decrypted temporarily in SvelteKit backend endpoints to route request headers safely. They are never exposed to client-side code.
We support standard provider endpoints. Most commonly, you
will target /v1/chat/completions (OpenAI, Gemini, DeepSeek, Amazon, etc.) and /v1/messages (Anthropic). Check our integration documentation recipes for
default request bodies.
During a proxied API request, our backend captures the incoming payload size and analyzes the response metadata (e.g. usage statistics returned by OpenAI/Anthropic). If usage metadata is absent, we run a fallback character count estimation. We then translate the prompt and completion tokens into exact USD costs based on official provider pricing sheets.
If the cumulative spending of a key or model exceeds the threshold you configured, our backend proxy will block any subsequent requests immediately with a clear 402 Cost Limit Exceeded error. This prevents accidental massive costs due to developer scripting loops.