Guides / gemini 2.5 pro pricing
Gemini 2.5 Pro pricing for long-context work
Gemini 2.5 Pro pricing is $1.25 per million input tokens and $10.00 per million output tokens, an eight-to-one spread. Cheap input makes long-context analysis viable, and expensive output punishes narration. Culpa, a local-first LLM cost, margin, and forecast ledger, prices each call from that rate and attributes it to the document run that spent it.
Why this happens
Pro is where teams put large document sets, because $1.25 per million input makes stuffing a very long context genuinely affordable. That same economics creates the trap. A hundred-thousand-token context is twelve and a half cents per call, which sounds free until it runs on every request in a batch. The output side then compounds it, because analysis tasks tend to produce long answers at eight times the input rate.
What this usually looks like
- A batch job over a document set costs more than your entire interactive product.
- Context size per call has grown steadily as more documents were added to the retrieval step.
- Nobody has measured how much of the retrieved context the model actually used.
- Output length on analysis calls has never been capped, because the answers looked useful.
Free, no card, no account
Run the free Cost Leak Scan
It shows your most expensive conversation before you install anything.
Mistakes that cost the most
| Mistake | Why it hurts | Do instead |
|---|---|---|
| Retrieving more context because input is cheap. | Cheap isn't free. A hundred thousand tokens is $0.125 a call, which is real money across a batch. | Measure retrieval precision. Cutting unused context is free quality and free money at once. |
| Letting analysis output run unbounded. | Output bills eight times input, so a long analysis can outweigh a very long context. | Cap max tokens and request structured output, then check whether anything downstream lost information. |
| Running a whole batch before pricing one item. | Batch jobs multiply a per-item cost nobody checked, and the bill arrives after the run. | Price one item end to end, multiply by the batch size, and approve that number before running. |
Run this check tonight
- Price a single batch item: context tokens times $1.25 per million plus output tokens times $10.00.
- Multiply by your batch size before running it. Compare that to what you expected.
- Measure what share of retrieved context the answer actually references.
- Check whether cached input is in use. At $0.125 per million it's a tenfold discount on repeated context.
- Compare your output-to-input token ratio against the 8x price ratio. That tells you which side to work on.
A document batch, priced before it runs
Illustrative example
A batch analysing 4,000 documents. Each call sends 80,000 tokens of context and returns 1,500 tokens of structured analysis, on Gemini 2.5 Pro.
$460 a run, and halving the retrieved context saves $200 of it. The output side is only 13% of the bill here, so context precision is the lever rather than answer length.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $1.25 / $10.00 per million | Gemini 2.5 Pro input and output rates | calculated | Culpa price book row for gemini/gemini-2.5-pro, $0.00125 and $0.010 per 1k, effective 2026-07-02, verified against Gemini API pricing. |
| $260 to $460 | modelled cost of one 4,000-document batch run, at halved versus full context | estimated | Both endpoints from the teardown arithmetic at real Gemini 2.5 Pro rates. A range because document count and context size are modelled. |
What a generic answer can’t know
A rate card can't tell you that half your retrieved context is never referenced in the answer, or that one batch job outspends your interactive product. That needs per-call token data tied to the job that produced it. Culpa records it on your infrastructure, keeps your prompts and responses there, and counts the calls to run your plan.
Questions founders ask next
How much does Gemini 2.5 Pro cost?
$1.25 per million input tokens and $10.00 per million output, verified 2026-07-02 against the Gemini API pricing page. Cached input is $0.125 per million, a tenfold discount that matters most on repeated long contexts.
Is long context actually affordable on Gemini 2.5 Pro?
Per call, yes. A hundred thousand tokens of context is about twelve and a half cents. Across a batch of thousands it stops being trivial, so price one item and multiply before you run the job rather than after.
Should I use Pro or Flash for document analysis?
Flash is roughly four times cheaper on both sides, so start there and move up only if quality demands it. The right test is your own documents, because the gap between the two models varies enormously by task.
On your infrastructure
Culpa runs on your infrastructure. Your prompts and responses never leave it. Culpa counts calls to run your plan, and it fails open, so if it ever breaks your app keeps running.
Free, no card, no account
Run the free Cost Leak Scan
It shows your most expensive conversation before you install anything.
Keep reading
Sources: Gemini API pricing. Last reviewed 2026-08-01, rates effective 2026-07-02. Plain text version.