AI Batch API Cost Calculator

The Batch API is an asynchronous endpoint for jobs that do not need an immediate reply. You upload many requests at once; the provider processes them within a completion window of up to 24 hours and returns the results in one file. In exchange for giving up real-time latency and streaming, OpenAI, Anthropic, and Google Gemini each charge a flat 50% discount on their standard synchronous per-token prices. That discount is a published, vendor-guaranteed term, so this tool applies one multiplier of 0.5 across all three providers.

Token costs use the standard LLM formula. Let I be the input tokens per request, O the output tokens per request, N the number of requests, Pin the standard input price and Pout the standard output price (in USD per 1,000,000 tokens):

1. inputCostStd = I ÷ 1,000,000 × Pin × N
2. outputCostStd = O ÷ 1,000,000 × Pout × N
3. standardTotal = inputCostStd + outputCostStd
4. batchTotal = standardTotal × 0.5 — the 50% batch discount
5. saving = standardTotal − batchTotal, which equals standardTotal × 0.5 — the saving is always exactly half the standard bill

The LKR figures are the USD results multiplied by an editable USD→LKR rate; they are a secondary convenience for local budgeting, not a live exchange feed. The calculator also cross-checks itself: the batch total is computed twice — once as half the standard total, and once by applying the halved per-token rates (Pin × 0.5, Pout × 0.5) directly to the token counts. Both methods agree to the cent, the same way the income-tax calculator reconciles two IRD formulas.

Input and output are priced separately because the model spends compute generating each output token, while input tokens are read once — output is almost always dearer. Because the discount is flat, the model you pick and the input/output mix change the totals but never the 50% ratio. The figures exclude prompt caching, image or audio tokens, and per-tenant contract pricing, which are handled by separate tools linked below.

• OpenAI — Batch API guide (50% discount, 24-hour window)
• OpenAI — API pricing (standard per-token rates)
• Anthropic — Message Batches API (50% discount on input & output)
• Anthropic — Pricing (standard per-token rates)
• Google — Gemini API Batch Mode (50% of interactive price)
• Google — Gemini API pricing (standard per-token rates)
• Central Bank of Sri Lanka — indicative USD→LKR rate

The 50% batch term is documented by all three providers and does not drift. Standard per-token prices are list prices that change without notice; they were last cross-checked on 2026-06-06. Confirm against each provider's current pricing page before relying on a number for a large spend.