Guide
AI API token cost calculator
Price a coding-agent workload here. The same calculator stays on the homepage. Change the model and monthly token mix, then keep the same client.
Create account Calculate savings
Price the workload first
Pick a model
Choose the plan and model your agent actually calls.
Set the token mix
Enter monthly input, output, cache read, and cache write tokens.
Use a preset
Start from light coding, daily agent, or heavy automation, then edit.
Compare listed vs official
The calculator uses live catalog rates next to public list prices.
What the calculator measures
Coding agents spend most of their money on tokens, not on seats. The useful question is not a vague “cheap AI API.” It is: what does this model cost for this input/output/cache mix?
This page runs the live calculator with no signup wall. It also stays on the homepage at /#savings. Both use the public catalog: listed cost, public list cost when we have one, and the save percentage.
How to enter usage
If you already have provider invoices or a Token Supply usage export, copy the monthly totals. If you are planning a new agent, start from a preset and adjust.
- Input tokens: prompts, repo context, and tool results you send.
- Output tokens: completions, diffs, and long reasoning traces.
- Cache read: reused prefix tokens on later turns.
- Cache write: tokens written into the prompt cache.
Related conversions on this page
One form answers two questions: what these tokens cost in dollars, and what the official list price is versus the Token Supply listed price. You do not need a second tool.
Read the result
Listed cost is what Token Supply charges for that mix. Official cost is the public provider list under standard-context assumptions. Save percentage is the gap.
Some models omit an official comparison when there is no public baseline. The listed cost is still valid.
After you pick a model, point Codex, Claude Code, OpenCode, Cursor, or an SDK at Token Supply with a prepaid key.
Reduce coding-agent API costs · Use Codex with a custom endpoint · Use Claude Code with a custom endpoint
Details
- Are the rates live?
- Yes. The calculator loads GET /api/v1/pricing and uses current listed rates.
- Does this include image models?
- No. Image generation is not in the current catalog.
- Can I cap spend after I switch?
- Yes. Token Supply is prepaid. Requests stop when the wallet reaches zero.
Related
- Reduce coding-agent API costs — Keep your client and lower token spend on the same workload.
- Track AI API token usage — Prepaid wallet, usage dashboard, and per-key spend.
- Codex custom endpoint — config.toml, auth.json, and the Responses API.
- Claude Code custom endpoint — ANTHROPIC_BASE_URL, auth token, and settings.json.
- OpenCode custom provider — OpenAI, Anthropic, Gemini, and Grok presets.