The OpenAI API, 40% cheaper.
Draftworks serves the same OpenAI models through the same API contract at lower prices: GPT-5.6 Sol at $2.40 per million input tokens and $12.00 per million output against OpenAI's $5.00 and $30.00. A signed receipt on every response proves which model answered.
| Per 1M tokens | OpenAI | Draftworks Base | Draftworks Scale |
|---|---|---|---|
| GPT-6 Astragpt-6-astra | |||
| Input | $10.00 | $6.00 | $6.00 |
| Cached input | $1.00 | $0.60 | $0.60 |
| Output | $50.00 | $30.00 | $30.00 |
| GPT-5.6 Solgpt-5.6-sol | |||
| Input | $4.00 | $2.40 | $2.40 |
| Cached input | $0.40 | $0.24 | $0.24 |
| Output | $20.00 | $12.00 | $12.00 |
| GPT-5.6 Terragpt-5.6-terra | |||
| Input | $2.00 | $1.20 | $1.20 |
| Cached input | $0.20 | $0.12 | $0.12 |
| Output | $12.00 | $7.20 | $7.20 |
| GPT-5.6 Lunagpt-5.6-luna | |||
| Input | $0.20 | $0.12 | $0.12 |
| Cached input | $0.02 | $0.012 | $0.012 |
| Output | $1.20 | $0.72 | $0.72 |
| GPT-5.4 Progpt-5.4-pro | |||
| Input | $30.00 | $18.00 | $18.00 |
| Output | $180 | $108 | $108 |
OpenAI column is OpenAI's published API price. Base applies from your first dollar; Scale applies permanently once lifetime purchases reach $5,000. The same discount covers every model in the catalog, including Grok 4.6 and Kimi K3.
Why the same tokens cost less.
Draftworks serves inference from Microsoft Azure's fleet, purchased at committed volume rather than retail. The models, weights, and API behavior are identical; the margin structure is not, and the difference funds the discount.
Cheaper resellers usually ask you to trust that they have not quietly swapped in a smaller model. Draftworks removes the trust question: every response is signed over the model id and the exact bytes served, receipts are retrievable forever, and an open audit method compares token distributions against OpenAI's own endpoint.
Common questions.
- How is Draftworks cheaper than the OpenAI API?
- The same models are served from Microsoft Azure's inference fleet, which we buy at committed volume. The unit economics differ from retail API pricing, and the difference is passed on: every model is 40% under the provider's published price on both Base and Scale.
- Are these the same models OpenAI serves?
- Yes. Each response reports the pinned model version, for example gpt-5.6-sol-2026-07-09, and carries an Ed25519-signed receipt over the model id, token counts, and hashes of the exact request and response bytes. You can verify any receipt cryptographically, and an open audit script compares logprob distributions against OpenAI's endpoint directly.
- What do I have to change in my code?
- One line: point your OpenAI SDK's base URL at https://api.draftworks.dev/v1 and use a Draftworks API key. Request shapes, response shapes, streaming, and tool calls are unchanged.
- Is there a subscription or minimum commitment?
- No. Billing is prepaid credits starting at $5. A credit is a dollar of inference at your tier's rates, on any model.
- Which models are available besides OpenAI's?
- Claude Fable 5.1, Fable 5, Opus 5, Sonnet 5, Grok 4.6, Kimi K3, and GLM 5.3 use the same 40% discount. Image generation and realtime voice are available on Scale.