TokenAtlasGemini pricing tiers
Gemini 1.5 Pro: $1.25 in / $5 out per 1M tokens (≤128K context). Above 128K: $2.50 in / $10 out. Flash: $0.075 in / $0.30 out — the cheapest frontier model on the market.
Multimodal cost
Images, video frames and audio seconds convert to tokens at fixed ratios. The calculator handles conversion automatically when you specify the input type.
Where Gemini wins
High-volume classification, multimodal RAG and 1M-token context windows. Flash routinely undercuts GPT-3.5 by 4-6×.
Worked example: a document Q&A workload
A workload answering questions over 50,000 monthly documents at 3,000 input / 200 output tokens each is 150M input and 10M output tokens. On Gemini 1.5 Pro (≤128K context): 150M/1M x $1.25 = $187.50, plus 10M/1M x $5 = $50, for $237.50/month. The same shape on Gemini 2.5 Pro at $1.25/$10 in/out is $187.50 + $100 = $287.50, reflecting the higher output rate.
Cost drivers ranked
Context length ranks first for Gemini workloads specifically, since crossing into a larger context tier or feeding in long documents changes the input token count more than any other input. Output length ranks second, call volume third. Multimodal inputs (images, audio) convert to tokens at fixed ratios and should be added to the input total before comparing against a text-only alternative.
Choosing between Gemini 1.5 Pro and 2.5 Pro
Blended at 70/30: Gemini 1.5 Pro = 0.7x$1.25 + 0.3x$5 = $2.375/1M tokens. Gemini 2.5 Pro = 0.7x$1.25 + 0.3x$10 = $3.875/1M. Input pricing is identical between the two; the entire gap sits in the output rate, so the choice matters most on workloads with long generated responses (reports, drafts) rather than short extractive answers pulled straight from retrieved context.
A second scenario at smaller scale
A team running 2,000 monthly Q&A calls at 3,000 input / 200 output tokens on Gemini 1.5 Pro (≤128K) is 6M input and 0.4M output tokens: 6/1M x $1.25 = $7.50, plus 0.4/1M x $5 = $2, for $9.50/month. Compare that to the 50,000-document workload on this page modelling to $237.50 — volume alone accounts for roughly a 25x gap between the two scenarios on the identical model.
Translating the estimate into cost per document answered
Dividing the monthly total by document volume gives a per-unit figure that's easier to compare against a manual-review baseline: $237.50 / 50,000 documents = $0.00475 per document on Gemini 1.5 Pro. That per-unit number is what to carry into a pricing or budgeting conversation, rather than the aggregate monthly figure, since it stays roughly constant as volume scales.
What this page does not do
This calculator prices token volumes you specify against Gemini's published per-1M rates; it does not pull usage from a Google Cloud billing account or apply Vertex AI enterprise discounts automatically. Multimodal conversion ratios (image, audio, video) are applied using fixed published factors, not measured from your actual files, so treat mixed-media estimates as directional.
Frequently asked questions
- Is the 1M context tier supported?
- Yes — both the standard and extended context price brackets are modelled.
- Can I compare Gemini against GPT and Claude?
- Yes — every calculator can be loaded side-by-side with the comparison view.
- Does it include Vertex AI pricing?
- Direct API pricing is default; Vertex enterprise rates are on the Team plan.