Loading…
Loading…
Your AI API bill on the day each price change lands, and the cheapest switch
Dated price change · Introductory prices with an end date
Peak-hour pricing · Double from 01:00-04:00 and 06:00-10:00 UTC on weekdays
No change announced · List prices checked 13 Oct 2026
Tier
Every prompt token sent, cached or not, in millions.
Count thinking tokens: Gemini 4 Argon bills them as output.
Billed at the cache-read price. Argon caches implicitly, so take the hit rate from your usage data.
Moves DeepSeek rates only, including DeepSeek in the comparison below.
Presets reuse the worked examples in our Gemini 4 Argon pricing breakdown and DeepSeek peak-hour guide, so the numbers here match theirs.
USD per 1M tokens, from each vendor’s own pricing page. Gemini and DeepSeek checked 18 Oct 2026, the rest 13 Oct 2026. Dates are inclusive. DeepSeek peak hours: 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday, Chinese public holidays excepted.
| Model | Tier | When | Input | Cache read | Output |
|---|---|---|---|---|---|
| Gemini 4 Argon | Standard | 13 Oct to 30 Nov 2026 | $2.00 | $0.10 | $10.00 |
| Gemini 4 Argon | Standard | From 1 Dec 2026 | $4.00 | $0.20 | $20.00 |
| Gemini 4 Argon | Batch | 13 Oct to 30 Nov 2026 | $1.00 | $0.05 | $5.00 |
| Gemini 4 Argon | Batch | From 1 Dec 2026 | $2.00 | $0.10 | $10.00 |
| Gemini 4 Argon | Flex | 13 Oct to 30 Nov 2026 | $1.00 | $0.05 | $5.00 |
| Gemini 4 Argon | Flex | From 1 Dec 2026 | $2.00 | $0.10 | $10.00 |
| Gemini 4 Argon | Priority | 13 Oct to 30 Nov 2026 | $3.60 | $0.18 | $18.00 |
| Gemini 4 Argon | Priority | From 1 Dec 2026 | $7.20 | $0.36 | $36.00 |
| Gemini 3.8 Flash | Standard | Until 31 Dec 2026 | $0.75 | $0.075 | $3.75 |
| Gemini 3.8 Flash | Standard | From 1 Jan 2027 | $1.50 | $0.15 | $7.50 |
| Gemini 3.8 Flash | Batch | Until 31 Dec 2026 | $0.375 | not modelled | $1.875 |
| Gemini 3.8 Flash | Batch | From 1 Jan 2027 | $0.75 | not modelled | $3.75 |
| DeepSeek V4.1 Flash | Standard | Off-peak hours | $0.15 | $0.003 | $0.60 |
| DeepSeek V4.1 Flash | Standard | Peak hours | $0.30 | $0.006 | $1.20 |
| DeepSeek V4 Pro | Standard | Off-peak hours | $0.66 | $0.022 | $1.98 |
| DeepSeek V4 Pro | Standard | Peak hours | $1.32 | $0.044 | $3.96 |
| Claude Fable 5.1 | Standard | No change announced | $10.00 | not modelled | $50.00 |
| GPT-6 Astra | Standard | No change announced | $10.00 | not modelled | $50.00 |
| Claude Opus 5.5 | Standard | No change announced | $4.00 | not modelled | $20.00 |
| Claude Sonnet 5.5 | Standard | No change announced | $2.00 | not modelled | $10.00 |
| GPT-6.1 Sol | Standard | No change announced | $2.00 | not modelled | $10.00 |
| Gemini 3.1 Pro Preview | Standard | No change announced | $2.00 | not modelled | $12.00 |
| Grok 4.7 | Standard | No change announced | $2.00 | not modelled | $6.00 |
| Claude Haiku 5.5 | Standard | No change announced | $0.10 | not modelled | $0.50 |
| GPT-6 Luna | Standard | No change announced | $0.10 | not modelled | $0.50 |
List prices only. Not modelled: cache writes and storage, long-context surcharges, Anthropic and OpenAI batch discounts, free tiers, taxes and volume deals. For one static price per model across request volumes, use the AI model cost calculator; for multi-step agent loops, the AI agent cost simulator. The arithmetic behind the defaults is in the Gemini 4 Argon price doubling post and the DeepSeek V4.1 Flash 1017 pricing guide.
Run your projects on Claude Code?
We packaged the .claude config that runs this site — 26 specialist agents, 14 workflow skills, and 6 rule files from 31 real production incidents. From $9.
Get the Claude Code Production Pack — $29Most AI API calculators show one price per model, but prices now come with dates. Gemini 4 Argon’s introductory rate ends on 30 November 2026 and every tier doubles the next day. Gemini 3.8 Flash follows on 1 January 2027, and DeepSeek charges double in peak hours every weekday. This calculator takes one month of your usage and prices it on each date the price changes. It shows the jump in dollars and percent, counts the days left, and ranks the three cheapest models at the same usage on the change date.
Every price is stored with a start date and an end date, both inclusive, plus the page it came from and the day it was checked. Gemini 4 Argon’s Standard tier is $2.00 input, $10.00 output and $0.10 cache read per 1M tokens from 13 October to 30 November 2026, then $4.00, $20.00 and $0.20 from 1 December. The calculator finds the price in force on your local date and every announced price after it.
The monthly bill is input tokens times the uncached share times the input price, plus input tokens times the cached share times the cache-read price, plus output tokens times the output price, with tokens counted in millions. At 400M input, 60% cached and 20M output on Standard, that is 160 × $2.00 + 240 × $0.10 + 20 × $10.00 = $544 now, and $1,088 from 1 December. Where no cache-read price is on record, cached tokens bill at the input price and the row says so.
DeepSeek has one rate card and two clocks. The calculator blends the off-peak and peak prices by your peak share: at 21%, deepseek-flash input costs 0.79 × $0.15 + 0.21 × $0.30 = $0.1815 per 1M. A team working 09:00 to 18:00 IST overlaps the peak windows for 4.5 of its 9 hours, which is where the 50% preset comes from.
The countdown reads your clock once when the page loads and counts calendar days to the change. Rivals are priced on the change date, on your tier where they offer it and on Standard otherwise, and any later change of their own is flagged.
A startup running a support bot on Gemini 4 Argon, deciding whether to move it to Batch or to another model before the 1 December doubling reaches the December invoice.
A finance lead forecasting Q1 2027 API spend, with Gemini 3.8 Flash doubling on 1 January and Gemini 4 Argon already on its new price.
A team in India on deepseek-flash, measuring how much of the bill comes from office-hours traffic that lands in the 11:30 to 15:30 IST peak window.
A developer posting a before-and-after bill card in the team channel to back a migration ticket, with the cheapest same-usage rival beside it.
Scope note: List prices only, checked 18 October 2026 for Gemini and DeepSeek and 13 October 2026 for the rest. Not modelled: cache writes and storage, long-context surcharges, Anthropic and OpenAI batch discounts, free tiers, taxes and volume deals. Google states no time zone for its cutover. The month is whatever usage you enter; a billing cycle that straddles the change date splits between both prices.
Pick the model you pay for. Gemini 4 Argon and Gemini 3.8 Flash have dated price changes, DeepSeek prices by time of day, and the rest have no change announced.
Choose the tier if the model has more than one: Standard, Batch, Flex or Priority on Argon, Standard or Batch on Flash.
Enter last month’s input and output tokens in millions and the share of input that was a cache hit. Count thinking tokens as output.
For DeepSeek, set the share of traffic in peak hours or pick a preset such as IST office hours (50%).
Read the bill before and after the change, the days left and the three cheapest models at the same usage, then share the result card.
About the AI API Price Change Calculator
On 1 December 2026. Google’s Gemini API pricing page lists the current rates through 30 November and double rates from 1 December, on Standard, Batch, Flex and Priority alike. Google gives the date but no time zone, so the calculator counts days on your local calendar and prices traffic on 1 December, your time, at the new rate.
By exactly 100% for the same tokens, because the input, output and cache-read prices all double. A chatbot sending 50M input and 10M output tokens a month goes from $200 to $400 on Standard. An agent with 400M input tokens (60% cached) and 20M output goes from $544 to $1,088. Batch halves either figure but does not avoid the doubling.
Yes. Google’s pricing page says the output price includes thinking tokens, so a reasoning-heavy request pays the output rate for tokens your users never see. Enter output as the billed total from your usage dashboard, thinking included, or the calculator will understate the bill.
DeepSeek charges double from 01:00 to 04:00 and from 06:00 to 10:00 UTC, Monday to Friday, except on Chinese public holidays: 35 of the 168 hours in a week. Off-peak, deepseek-flash costs $0.15 per 1M cache-miss input tokens, $0.003 per 1M cache hits and $0.60 per 1M output; at peak it is $0.30, $0.006 and $1.20. The calculator blends the two by your peak share: 21% for round-the-clock traffic, 50% for a team in India working 09:00 to 18:00 IST.
It depends on your token mix, so the calculator ranks every model at your usage under the prices in force on the change date. For an agent at 400M input (60% cached) and 20M output, DeepSeek V4.1 Flash at 21% peak traffic comes to $44.43 a month on 1 December, against $1,088 on Gemini 4 Argon Standard. Price per token says nothing about quality, so test a cheaper model on your own tasks before you switch.
The static price table behind this tool lists input and output prices for the Claude, GPT-6, Grok and Gemini 3.1 Pro Preview models but not their cache-read rates, so cached tokens on those models bill at the full input price here and their real bill with caching is lower. Gemini 3.8 Flash Batch has the same gap, because Google lists no Batch cache-read price.
No. The calculator is JavaScript running in your browser and calls no server. The share link carries numbers only (a model code, a tier code and your token counts), and the share image shows bills, percentages and the day count.
Find what breaks when Chrome 158 turns XSLT off on 17 Nov 2026
Open →DeveloperHow many days your $100, $200 or $500 API credit lasts
Open →DeveloperScore your system prompt for bloat and get a cut list
Open →DeveloperPrice an agent fleet per run, day and month on every major model at once
Open →