Batch API Savings Calculator
See how much the Batch API saves on a large LLM job: the same requests processed asynchronously at a discount (50% at OpenAI, Anthropic and Google), with the batch cost per request.
Last updated: October 2026
Saved with the Batch API
$160.00
Formula below · 3 sources (developers.openai.com, platform.claude.com, ai.google.dev) · Updated Oct 2026
Compare with similar
About this calculator
Batch APIs take a file or list of requests, process them asynchronously and return the results later, in exchange for a lower price. As of October 2026, OpenAI’s Batch API, Anthropic’s Message Batches API and the Gemini Batch API all charge 50% of the standard price for both input and output tokens.
standard cost = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000 batch cost = standard cost × (1 − discount)
Batches are meant for work that can wait: evaluations, classification and tagging of large datasets, document summaries, embeddings back-fills. Each provider targets completion within 24 hours, and most batches finish sooner. The discount stacks with prompt caching. The model list uses the standard list prices; long-prompt rates (GPT-6 over 272,000 input tokens, Claude Haiku 5.5 over 100,000) are applied before the discount.
| Model | Input | Output | Note |
|---|---|---|---|
| GPT-6 Astra | $10 | $50 | $20 / $75 over 272K-token prompts |
| GPT-6.1 Sol | $2 | $10 | $4 / $15 over 272K-token prompts |
| GPT-6 Luna | $0.10 | $0.50 | $0.20 / $0.75 over 272K-token prompts |
| Claude Fable 5.1 | $10 | $50 | |
| Claude Opus 5.5 | $4 | $20 | |
| Claude Sonnet 5.5 | $2 | $10 | |
| Claude Haiku 5.5 | $0.10 | $0.50 | $0.50 / $2.50 over 100K-token prompts |
| Gemini 4 Argon | $2 | $10 | to Nov 30, 2026 |
| Gemini 4 Argon | $4 | $20 | from Dec 1, 2026 |
| Gemini 3.8 Flash | $0.75 | $3.75 | to Dec 31, 2026 |
| Gemini 3.8 Flash | $1.50 | $7.50 | from Jan 1, 2027 |
| Gemini 3.5 Flash-Lite | $0.30 | $2.50 |
How to use
Choose the model, enter the average input and output tokens per request, the number of requests and the provider’s batch discount (50% for all three as of October 2026). Example (the defaults): 10,000 requests of 3,000 input and 1,000 output tokens on Claude Opus 5.5 ($4 / $20). Standard: 10,000 × (3,000 × 4 + 1,000 × 20) ÷ 1M = $320.00. Batch at 50%: $160.00, saving $160.00, or $0.016 per request. Example 2: the same job on GPT-6 Astra ($10 / $50): standard 10,000 × (30,000 + 50,000) ÷ 1M = $800, batch $400, saving $400. Example 3: on Gemini 3.5 Flash-Lite ($0.30 / $2.50): standard $34.00, batch $17.00.
Frequently asked questions
How long does a batch take?
OpenAI, Anthropic and Google process batches within a 24-hour window; many finish much sooner, depending on size and load. Use the regular API for anything a user is waiting for.
Can I combine batch and prompt caching?
Yes. Anthropic states that the Batch API and prompt-caching discounts stack, and OpenAI’s batch price list includes cached-input rates. Cache hits are less predictable in a batch because requests may run in any order.
What kinds of jobs suit the Batch API?
Anything offline and repeatable: running evaluations, classifying or extracting data from large sets of documents, generating product descriptions, translating catalogues, or re-processing a backlog. Interactive chat and anything with a deadline under a day should use the standard API.
How current are the prices?
They are the standard list prices on the OpenAI, Anthropic and Google pricing pages as of October 2026 (checked 17 October 2026), before taxes, volume discounts or regional surcharges. Google has announced increases for Gemini 4 Argon (from December 1, 2026) and Gemini 3.8 Flash (from January 1, 2027); both prices are in the list. Providers change prices often, so confirm on the official pricing page before you commit a budget.
Related calculators
Sources & references
Think a result is wrong? Report a problem with this calculator.