Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesAI API bills usually separate the tokens you send from the tokens a model generates. Eligible repeated input may be charged at a cached-input rate, while some caching features also charge for writes or storage. To estimate a request, use the exact model, usage mode, and rate categories on the provider’s current pricing page, then check the usage reported by the API.
What you pay for in an AI API request
Providers generally meter input and output separately. Input tokens are the material sent to the model; output tokens are generated by it. A provider’s pricing table may split these further into uncached input, cached input, cache writes, and output. The categories and billing rules are provider-specific, so do not assume a rate or cache rule from one API applies to another. OpenAI’s current pricing page, for example, lists distinct token categories and may show separate rates for context lengths or other configurations.
As an Amazon Associate I earn from qualifying purchases.
Generated reasoning tokens can count as output even if they are not visible in the final answer. As a result, visible response length alone may not tell you how much output was billed. Check the model’s usage reporting and pricing definitions; OpenAI’s token guidance explains this distinction.
A practical cost equation
Use this as a planning framework, then confirm the provider’s definitions and billing records:
#1 Best Overall
- Used Book in Good Condition
Estimated token charge = uncached input tokens × uncached-input rate + cached input tokens × cached-input rate + output tokens × output rate + applicable cache-write, cache-storage, or feature charges.
If a price is listed per million tokens, divide the token count by 1,000,000 before multiplying by that rate. Do not automatically add a separate cache-write fee: OpenAI’s current guide presents cache-write pricing as an alternative input-token rate, not an additive fee.
For a hypothetical request with 10,000 input tokens and 1,000 output tokens, multiply each count by the selected model’s respective per-token rate. For an actual estimate, split eligible cached input from uncached input and include any applicable storage or non-token charges. The example shows the arithmetic only; it is not a quoted bill.
Recommended Free Tools
Why tokens do not map neatly to words
Tokens are the units an API uses to process text, not a fixed word count. OpenAI’s rough English-language estimates are about 4 characters per token, three-quarters of a word per token, or 75 words per 100 tokens. These are approximate examples, not conversion constants: language, spelling, capitalization, spaces, and the model’s tokenizer all affect the count.
A plain-text estimate can also miss message structure, tool definitions, schemas, images, and files. For a more dependable estimate, use model-specific tokenization where available and compare it with the usage figures returned by the API.
How caching changes the calculation
Caching can lower the rate for eligible repeated prompt content, but the mechanics matter: what must repeat, how much content qualifies, how long it remains cached, and whether writes or storage incur charges. Do not treat “cached” as synonymous with free.
OpenAI prompt caching
OpenAI says the rendered prompt prefix must match for cache reuse, and eligible lengths and breakpoints depend on the model. Its current guide specifies a minimum cacheable prompt length of 1,024 tokens for GPT-5.6 and later models; thresholds vary for earlier models. These are model-specific rules and can change, so check the current prompt-caching guide for the model you intend to use.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallAs an illustration of the rates documented for the named GPT-5.6-and-later models, OpenAI says cache writes cost 1.25 times the standard uncached-input rate; cache reads cost 0.1 times that rate for most of those models and 0.05 times for GPT-6.1 Sol. At a 0.1-times read rate, one write plus nine full reads would cost 2.15 times the ordinary input cost of one processing pass, rather than 10 times for ten uncached passes. This comparison assumes those published rates and full eligible reuse; it is not a universal caching result.
Google Gemini caching
Google describes implicit caching for Gemini 2.5 and newer models, alongside explicit caching as a separate feature. For explicit caching, cost depends on token count and time-to-live (TTL); the documented default TTL is one hour when unset, and storage duration can contribute to cost. Cached-token, uncached-input, and output charges may all apply. Google’s caching guide labels explicit caching Beta and describes endpoints and SDK methods under v1beta; check the current documentation for availability and implementation details.
Google notes that “At certain volumes, using cached tokens is lower cost than passing in the same corpus of tokens repeatedly.” The qualification matters: savings depend on volume, reuse, and the applicable cache and storage charges.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to compare API prices without misleading yourself
A price per million tokens is only one input to the total. Compare the model and configuration you would actually use, with a representative workload, rather than comparing provider averages or a single headline rate.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →- Match the task to a model. Compare options capable of doing the same work, not just models with similar names or advertised rates.
- Estimate the input and output mix. Include prompt size, expected completion size, and reported reasoning usage where available. Do not infer billed output from the visible answer alone.
- Check caching rules. Establish whether caching is implicit or explicit, what content must match, minimum-size and lifetime rules, read and write rates, and any storage fees.
- Match configuration and modality. Verify the service tier, context length, and whether the request uses text, image, audio, video, batch, priority, grounding, or another separately priced feature.
- Measure representative requests. Run sample tasks, inspect API-reported usage, and calculate cost per completed task at your expected volume. Repeat for different prompt shapes if your workload varies.
OpenAI’s Help Center cautions that “A lower price per million tokens does not necessarily produce a lower total cost: models can tokenize the same text differently and generate different amounts of output or reasoning.” It recommends: “Test representative tasks rather than comparing only the visible response length.”
Official price examples are snapshots, not market rates
Provider prices change, and a listed rate is specific to a model, tier, modality, and effective period. Recheck the relevant row before budgeting or deployment. OpenAI’s pricing page presents rates per one million tokens and distinguishes input, cached input, cache writes, and output; some models also have short- and long-context columns.
As one dated Google example, its pricing page listed Gemini 3.1 Flash-Lite Standard at $0.25 per 1 million text, image, or video input tokens; $0.50 per 1 million audio input tokens; $1.50 per 1 million output tokens; and $0.025 per 1 million text, image, or video cached tokens, plus $1.00 per 1 million tokens per hour for storage. These are Google AI for Developers figures accessed on October 7, 2026, not a general rate or a promise of current pricing. The same page lists different rates for Batch, Flex, and Priority tiers. See the current Gemini API pricing for the live model and configuration details.
Some Google listings also show scheduled rate changes. If a price is marked “through December 31, 2026” or “starting January 1, 2027,” keep the model, tier, modality, and effective date attached to the rate; prices from different periods are not interchangeable.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
