Gemini 2.5 Flash pricing at a glance
When you are already past the basics of API routing, the useful question is not simply which model to pick. It is how the model behaves across a full working session, where input tokens, output tokens and cache reads all add up. SwiftRelay makes that arithmetic visible: you pay per token, and every call deducts only what it used from the balance you topped up.
For Gemini 2.5 Flash, the current SwiftRelay pricing page lists the following rates per 1M tokens:
- Input: $0.1875
- Output: $1.50
- Cache read: $0.0194
That structure gives experienced users something practical to work with. Input tokens cover the material you send to the model, such as source files, prompts and context. Output tokens cover what the model writes back. Cache reads cover reuse of context that has already been processed. Because these are separate line items, you can see which part of a session is driving the bill instead of guessing at a flat monthly figure that may not match your actual usage.
There is no monthly fee and nothing to subscribe to with SwiftRelay. You top up a balance, and each API call deducts the amount it actually used. That makes the pricing page a working instrument rather than a background detail: you can estimate a session, run it, and then check the usage record, which lists tokens and cost per call.
A worked example of a steady coding day
Consider this example: a full day of coding with Gemini 2.5 Flash through SwiftRelay, using 200,000 input tokens, 40,000 output tokens and 100,000 cache-read tokens.
- Input cost: 0.2 × $0.1875 = $0.0375
- Output cost: 0.04 × $1.50 = $0.06
- Cache read cost: 0.1 × $0.0194 = $0.00194
- Example total: $0.09944
This is an example, not a promise about what your day will cost. The point is that the total is built from the published per-token rates, so you can adjust the inputs and see the result move. If your day involves more file context, the input figure changes. If you ask for longer explanations or larger patches, the output figure changes. If your workflow reuses a large cached context, the cache-read line keeps the calculation honest.
For a developer moving past the basics, this is a more useful way to plan than treating every request as identical. SwiftRelay records tokens and cost per call, which lets you compare a morning session against a heavier afternoon session and identify the calls worth reviewing. The balance approach also means there is no commitment to a recurring charge sitting behind your work.
A heavier example with larger context and cache reuse
Now consider a second example: a busy day that consumes 1,000,000 input tokens, 250,000 output tokens and 500,000 cache-read tokens.
- Input cost: 1 × $0.1875 = $0.1875
- Output cost: 0.25 × $1.50 = $0.375
- Cache read cost: 0.5 × $0.0194 = $0.0097
- Example total: $0.5722
Again, this is an example built directly from the SwiftRelay pricing page for Gemini 2.5 Flash. It shows why advanced users should watch the three categories separately. Output remains the largest part of this example, but input dominates many repository-scale workflows, while cache reads can become a meaningful line when a long context is reused repeatedly across related requests.
You do not need to guess which of your sessions deserve attention. SwiftRelay shows tokens and cost per call in the usage view, so you can spot a call that consumed far more output than expected, or a workflow that repeatedly sends context which could be handled differently. That visibility is the practical advantage of pay-per-token billing for people who already understand how their code and prompts behave.
Connect Gemini CLI with three copy-and-run lines
Advanced routing is only useful if the connection is quick. SwiftRelay is designed to get you running in minutes: copy three lines, set your key, and start your client. For Gemini 2.5 Flash, the model uses its native protocol through Gemini CLI, and the setup is as follows.
01 export GOOGLE_GEMINI_BASE_URL=https://api.fastapirelay.com 02 export GEMINI_API_KEY=<your swft_ key>
gemini
That is the whole path from an empty shell to a working Gemini CLI session pointed at SwiftRelay. There is no monthly fee to activate and no subscription to maintain. You top up first, then each call deducts only what it used.
Once you are inside the client, you can return to the SwiftRelay pricing page and compare Gemini 2.5 Flash with the other available models. Gemini 3.1 Pro is listed at $1.35 input, $7.81 output and $0.1347 cache read per 1M tokens. Gemini 3.8 Flash is listed at $0.9931 input, $4.46 output and $0.0986 cache read. Gemini 4 Argon is listed at $1.31 input, $6.28 output and $0.0653 cache read. Having the rates side by side makes it easier to decide when a heavier model is worth its cost for a specific task.
Managing balance, keys and per-call visibility
For an experienced workflow, control matters more than novelty. SwiftRelay supports up to 20 live API keys per account, which gives you room to separate projects, clients or environments without juggling a separate account for each one. Each key is still tied to the same transparent per-token billing model, and the usage view continues to list tokens and cost per call.
Top-up amounts are $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89. You choose the amount that fits your working rhythm, and your balance moves down as calls are made. If you top up and change your mind, unused credit can be refunded within 11 days of the top-up, minus what was used and the payment fee. That is a straightforward way to test a heavier workflow without committing to a fixed recurring arrangement.
Support is available through Messages in the dashboard, with login required. There is no support email to hunt for, which keeps the help path inside the same place where you review usage, manage keys and watch your balance. For a developer who wants answers and numbers in one place, that shortens the distance between a question and the next run.
Referral rewards that stay inside the same balance
SwiftRelay also rewards people who bring their own teams. When someone you invited tops up, you receive 14% of each top-up. When the people they invite top up, you receive 3% from those top-ups. Rewards can be moved into your balance for immediate use, or withdrawn once you reach the $4.86 minimum withdrawal amount.
New accounts also receive $0.22 in sign-up gift credit, which is enough to make a first call and see how the usage record works. That is the practical starting point: connect Gemini CLI, run a small task, check the tokens and cost per call, and then decide how you want to scale your day of coding on Gemini 2.5 Flash through SwiftRelay.
Your balance is topped up, your keys are ready, and your next coding session is one clean first run away — sign up for SwiftRelay now and get moving.