Gemini 2.5 Pro: 200K Context Pricing, Batch Costs and Old Preview IDs

Do not substitute a preview SKU

Google identifies stable gemini-2.5-pro separately from dated 2025 preview variants. Its deprecation page lists the stable release as 17 June 2025 and names retired preview IDs and their shutdown dates. A preview label or archive tutorial is not evidence that the same endpoint was still available in October 2026. Preserve the exact API model string, release type and cutoff date; a billing record for the stable model should not be merged with a retired preview.

The request-length threshold matters

The official paid API pricing lists up to 200,000 prompt tokens at $1.25 per million input and $10 per million output, with a different band above that threshold ($2.50 input and $15 output). The documentation also notes that output pricing includes thinking tokens. Treat the threshold as a request-level tier choice where specified, not a sliding tax on only the final few prompt tokens. If cached context or extra modalities are present, include separately documented handling.

Batch is a different operational bargain

At the short-context band, the Batch listing is $0.625 per million input tokens and $5 per million output tokens; longer context has its own batch band. A batch discount has practical constraints: asynchronous turnaround, queueing, potential noncompletion and result retrieval. It is not automatically applicable to every interactive application. Forecast separate interactive and asynchronous workloads, and never assign batch prices to an online request merely because the content is similar.

Practical planning example

A hypothetical request mix totaling 2 million short-context input tokens and 0.5 million output tokens would cost 2×$1.25 + 0.5×$10 = $7.50 at standard short-context list rates. The same counts under the applicable batch band would be $3.75, excluding storage and any ancillary billed service. The arithmetic assumes every request stays within the documented lower prompt-length tier. Running a few very long prompts can invalidate that simplified estimate.

Before choosing a paid tier

Obtain the billing quote for the exact Google product, account geography, tier, model or subscription, currency and commitment length. If the provider offers several kinds of meters, forecast each independently rather than summing quantities in incompatible units. The relevant captured meters here include input_tokens_per_1m, output_tokens_per_1m. Build a low, expected and high workload from your own request logs or editorial schedule; a pricing page is not a credible estimate of how many retries or approvals your team will need.

For governance, add an approval threshold for paid overages, reconcile invoices with metered activity, and schedule a quarterly or release-triggered source audit. The absence of an identified enterprise rate, promotional expiry or regional tax line is not evidence that the amount is zero. This guide does not represent a product trial, capability benchmark, procurement offer or warranty of uninterrupted access.

Historical availability and boundaries

Preserve the precise source version, checked date and any stated effective date before reusing these figures. A model launch announcement may confirm an introductory token rate without proving which private offers remain available; a mutable plan page can describe a later product state. If a source changes after this cutoff, publish a new snapshot instead of silently editing the historical record. Specialised product evaluations belong to the relevant AI 02–09 editorial teams, while this article documents the billing model and its limitations.