Account

Usage stats

API Keys

🤖 Dollar Store Tokens

Frontier LLMs at a fraction of the official price, with native endpoints for OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages. No signup, no KYC. You pay in Monero.

Price comparison

Per-request cost for 200K input and 50K output tokens, against official API pricing.

Model Official API Dollar Store Tokens Savings
claude-fable-5 $4.50 $2.84 37%
gpt-6-astra $4.50 $0.69 85%
claude-fable-5.1 $4.50 $2.84 37%
gpt-5.6-sol $2.50 $0.28 89%
gpt-5.5 $2.50 $0.38 85%

See all models →

One Price

One price per model. No tiers or volume brackets. The number you see is the number you pay, with no reading the fine print to figure out what you'll actually be charged. We handle routing across upstreams and guarantee the best price/performance ratio available, so you never have to pick routes or tune knobs. It just works.

Prompt caching is included. If we have to switch upstreams mid-conversation and your cached prefix goes cold on the new route, we eat the cost. You're billed at the cache-read rate for the tokens we expected to hit, not the full input rate. Our routing, our problem.

No Rate Limits

No concurrency caps or requests-per-minute throttling. No “upgrade for more” tiered throughput. Dollar Store Tokens is globally distributed across many servers, so paid requests scale as far as the hardware will carry them. We serve until the servers go down. The only thing that stops a legitimate paid request is capacity, not an artificial limit.

Real Models

Some proxy services advertise frontier models and quietly route you to a cheaper one. We don't. When you request a model, we route to that model, and if no healthy upstream is available for it, the request fails rather than silently substituting something else. You get the model you asked for, or you get an error.

Privacy

You pay in Monero. No name or email, no identity verification. A random token in a cookie is your account. We track token counts for billing, nothing more. Your requests pass through third-party infrastructure we don't control, so assume your prompts may be visible to unknown parties. We won't sell you a false promise on that. See privacy and terms.