For the ones
still up.
One API key for leading AI models, through OpenAI- and Anthropic-compatible endpoints. A fresh allowance every day, one flat monthly price, and no overage charges.
Catalog
Explore the model catalog
Browse the models available now. One API key, unified billing, and transparent rate multipliers.
How it works
Connect in three steps
Works with any app or SDK that lets you set an OpenAI- or Anthropic-compatible base URL.
Create your account
Sign up with Google or email and start on the free tier. No card needed.
Grab one API key
Create a single key in your dashboard that works across every model family.
Paste the base URL
Drop an endpoint below into your app or SDK and pick any model ID.
https://somny.xyz/v1https://somny.xyzSupports Chat Completions, Responses, and Messages · Connection guide
Daily allowance
A fresh allowance every day
Your daily token bucket refills automatically at midnight UTC—shared across every app and model.
Refills at 00:00 UTC
Every day starts with a full bucket shared across your apps and models. Unused tokens don’t roll over.
Weighted by model rate
Each model has a rate multiplier: 1,000 tokens at 2× use 2,000 tokens of your allowance. See every model’s rates.
Hard cap, $0 overages
Requests pause cleanly when you reach 100% until the next refill. Never an automatic top-up or surprise charge.
Daily Capacity
Simple monthly plans
Pick the daily allowance that fits your pace. Flat monthly pricing with zero surprise overage fees.
Questions about rates or which plan fits your workflow? Reach out to me on Discord.
Some models use a higher rate for very long prompts. Model rates · How usage is counted
Questions
Frequently asked
What are weighted tokens?
Each model has multipliers for input, output and cached tokens. A request’s actual tokens are multiplied by those rates and taken from your daily allowance, so 1,000 output tokens at 2× use 2,000 weighted tokens. The Models page lists every rate.
What happens when I reach my daily limit?
New requests are declined until your allowance refills at midnight UTC. Nothing is charged automatically, and unused tokens don’t carry over.
Which apps can I use?
Apps and SDKs that let you set a custom OpenAI-compatible or Anthropic-compatible base URL. Somny supports Chat Completions, Responses and Messages for text and function tools. Image input, PDF input and structured output are available on marked models and endpoints. Hosted tools are unavailable, and features depend on the model and app. See the connection guide.
Does Somny store my prompts or responses?
No. Somny keeps usage metadata such as request IDs, token counts and timing, and processes request content only to fulfil each request. Model providers process requests under their own policies, and your app may keep its own history. Read the privacy policy.
Do long prompts cost more?
Some models use a higher rate above a long-context threshold. Where that applies, the threshold and rate are listed on the Models page and in your dashboard.
Can I cancel or get a refund?
Paid access lasts for the period you paid for. Stripe subscriptions renew monthly until you cancel them in Billing; manually arranged payments don’t renew. Refund requests are reviewed case by case under the refund and cancellation policy.
How do I get help?
Join the Somny Discord. When reporting an API problem, include the request ID from the x-request-id response header, and never post API keys in public channels.
Make room for your next idea.
Start on the free plan. No card needed.