Same models.
Smaller bill.

Claude, GPT, DeepSeek and more through one OpenAI-compatible API, for up to 50% less per token.

Get an API keyCompare prices

Add $5, get $5 more on your first top-up. Pay per token, no subscription.

Models fromAnthropicOpenAIDeepSeekGoogleMetaAlibabaxAIAion LabsByteDanceMiniMaxMoonshotXiaomiZ.aiand more

Works with the tools you already use

See setup guides
Claude CodeCoding agentCodex CLICoding agentCursorEditorOpenCodeCoding agentHermes AgentAgentKilo CodeCoding agent
ZedEditorOpenAI SDKSDKVercel AI SDKSDKLangChainFrameworkn8nAutomationOpen WebUIChat app

Prices, side by side.

Per million tokens. Our price on top, the model maker's list price struck through underneath.

Price per million tokens, ours and the list price
ModelInput, oursInput, listOutput, oursOutput, listDiscount
Claude Opus 5.5$2.68$4.00$13.40$2033%
Claude Sonnet 5.5$1.32$2.00$6.60$1034%
Claude Fable 5.1$6.50$10$32.50$5035%
GPT-6 Luna$0.066$0.10$0.33$0.5034%
DeepSeek V4.1 Flash$0.15$0.30$0.60$1.2050%
Claude Haiku 5.5$0.07$0.10$0.35$0.5030%
GLM-5.3 Flash$0.0825$0.15$0.275$0.5045%
Claude Sonnet 5$1.26$2.00$6.30$1037%
DeepSeek V4 Flash$0.174$0.30$0.696$1.2042%

Illustrative rates while we're in early access. Pricing never goes above the model maker's list price.

See every model

See what you'd save.

Pick the model you use most and what you spend on it a month at list price.

$2,000
$100$20,000
With PromptsForLess
$1,340 a month
You keep
$7,920 a year

At 33% below list price for Claude Opus 5.5.

How it's cheaper.

No lookalike models and no quality trade-off. The discount comes from AI credits that would otherwise go to waste.

  1. 1

    Teams buy more credits than they use

    Companies commit to AI credits up front and often end the month with plenty left over, credits that would otherwise go unused.

  2. 2

    We buy the spare credits and pool them

    We pay for that unused capacity at a discount and pool it into one supply that covers many models.

  3. 3

    You pay less for the same model

    Your requests run on the real model you asked for, billed per token below its list price.

Change one line.

Keep your SDK, your prompts and your code. Point the base URL at us and the savings start on the next request. Streaming, tool calls and structured outputs work as before.

from openai import OpenAI

client = OpenAI(
-    base_url="https://api.openai.com/v1",+    base_url="https://api.promptsforless.com/v1",    api_key=os.environ["PFL_API_KEY"],
)

client.chat.completions.create(model="claude-opus-5-5", ...)

Questions, answered.

Are these the same models?

Yes. Your request runs on the exact model you name, from the company that makes it. There are no lookalikes, distilled copies or swapped-in substitutes.

How can it be cheaper?

Companies often have AI credits they will never use. We buy that spare capacity at a discount, pool it, and pass most of the saving on to you.

Will I ever pay more than list price?

No. Every model is priced below its maker's list price, and you only pay for the tokens you use.

What is the first top-up offer?

We match your first $5 top-up, so you start with $10 of credit. It applies once per account.

Do you store my prompts?

We keep usage and billing records, such as token counts and timestamps. We do not keep the contents of your prompts or responses.

Which tools work with it?

Anything that speaks the OpenAI API: the OpenAI SDKs, the Vercel AI SDK, LangChain, LlamaIndex, and coding tools that let you set a custom base URL.

Got AI credits you won't use?

Sell them to us instead of letting them expire. Tell us which provider and roughly how much is left, and we'll get back to you with an offer.

  • Paid for unused credits, not left to expire
  • We review every submission by hand
  • Your details are only used to reply to you

Your next prompt costs less.

We match your first $5 top-up, so you start with $10 of credit. Setup takes about a minute.

Get an API key