Research · Provider pricing pages
LLM API pricing tracker in USD, CAD and JPY
As of October 14, 2026, standard API prices for the 15 current models tracked here range from $0.10 to $10.00 per 1M input tokens and $0.50 to $50.00 per 1M output tokens, according to each provider's own pricing page. At the Bank of Canada rate for October 13, 2026 (US$1 = C$1.4222, about ¥158.37), that is C$0.142 to C$14.22 (¥15.8 to ¥1,584) per 1M input tokens. Google has announced that Gemini 4 Argon prices double on December 1, 2026, and Gemini 3.8 Flash and 3.6 Flash prices double on January 1, 2027.
- Prices last checked
- Update cadence
- Checked weekly; prices change often, so check the date before quoting
- Exchange rates
- Bank of Canada daily average, October 13, 2026
- First published
- Compiled by
- Innovation Trigger Lab
Key figures
15
current models tracked from 5 providers: OpenAI, Anthropic, Google, Mistral AI and xAI
Prices checked October 14, 2026
C$1.4222
per US dollar, Bank of Canada daily average for October 13, 2026
JPY/CAD: 0.008980
¥158.37
per US dollar, derived through the CAD cross rate
1.4222 ÷ 0.008980
December 1, 2026
Gemini 4 Argon API prices double (input $2.00 → $4.00, output $10.00 → $20.00 per 1M tokens)
Gemini 3.8 Flash and 3.6 Flash double on January 1, 2027
API prices per 1M tokens in USD, CAD and JPY
Standard (pay-as-you-go, paid tier) prices for text input and output. USD is copied from each provider's pricing page; CAD and JPY are converted at the Bank of Canada rate for October 13, 2026. Models are grouped by provider in the order each provider lists them, not ranked.
| Provider and model | Input (USD) | Output (USD) | Input (CAD) | Output (CAD) | Input (JPY) | Output (JPY) |
|---|---|---|---|---|---|---|
| OpenAI GPT-6 Astra | $10.00 | $50.00 | C$14.22 | C$71.11 | ¥1,584 | ¥7,919 |
| OpenAI GPT-6.1 Sol | $2.00 | $10.00 | C$2.84 | C$14.22 | ¥317 | ¥1,584 |
| OpenAI GPT-6 Luna | $0.10 | $0.50 | C$0.142 | C$0.711 | ¥15.8 | ¥79.2 |
| Anthropic Claude Fable 5.1 | $10.00 | $50.00 | C$14.22 | C$71.11 | ¥1,584 | ¥7,919 |
| Anthropic Claude Opus 5.5 | $4.00 | $20.00 | C$5.69 | C$28.44 | ¥633 | ¥3,167 |
| Anthropic Claude Sonnet 5.5 | $2.00 | $10.00 | C$2.84 | C$14.22 | ¥317 | ¥1,584 |
| Anthropic Claude Haiku 5.5 | $0.10 | $0.50 | C$0.142 | C$0.711 | ¥15.8 | ¥79.2 |
| Google Gemini 4 Argon | $2.00 | $10.00 | C$2.84 | C$14.22 | ¥317 | ¥1,584 |
| Google Gemini 3.8 Flash | $0.75 | $3.75 | C$1.07 | C$5.33 | ¥119 | ¥594 |
| Google Gemini 3.5 Flash-Lite | $0.30 | $2.50 | C$0.427 | C$3.56 | ¥47.5 | ¥396 |
| Mistral Large 4 | $0.68 | $2.09 | C$0.967 | C$2.97 | ¥108 | ¥331 |
| Mistral Medium 3.5 | $1.50 | $7.50 | C$2.13 | C$10.67 | ¥238 | ¥1,188 |
| Mistral Small 4 | $0.15 | $0.60 | C$0.213 | C$0.853 | ¥23.8 | ¥95.0 |
| xAI Grok 4.7 | $2.00 | $6.00 | C$2.84 | C$8.53 | ¥317 | ¥950 |
| xAI Grok 4.3 | $1.25 | $2.50 | C$1.78 | C$3.56 | ¥198 | ¥396 |
Prices exclude sales tax, data-residency or regional-endpoint uplifts, priority or fast modes, and tool fees such as web search. Gemini output prices include thinking tokens. Converted values are rounded; use the USD figure for billing.
Notes on these prices
- Google Gemini 4 Argon: Price applies through November 30, 2026; see scheduled changes.
- Google Gemini 3.8 Flash: Price applies through December 31, 2026; see scheduled changes.
- Mistral Large 4: Sale price; regular price $1.36 input, $0.14 cached input, $4.18 output. The page gives no end date for the sale.
Cached input, batch and long-context prices
Most providers charge less for input tokens served from a prompt cache and for asynchronous batch jobs, and some charge more for long prompts. Values are USD per 1M tokens as listed; a dash means the page does not list a per-model price.
| Provider and model | Cached input | Batch input | Batch output | Long prompt: input / output |
|---|---|---|---|---|
| OpenAI GPT-6 Astra | $1.00 | $5.00 | $25.00 | >272K: $20.00 / $75.00 |
| OpenAI GPT-6.1 Sol | $0.10 | $1.00 | $5.00 | >272K: $4.00 / $15.00 |
| OpenAI GPT-6 Luna | $0.01 | $0.05 | $0.25 | >272K: $0.20 / $0.75 |
| Anthropic Claude Fable 5.1 | $0.25 | $5.00 | $25.00 | No tier listed |
| Anthropic Claude Opus 5.5 | $0.20 | $2.00 | $10.00 | No tier listed |
| Anthropic Claude Sonnet 5.5 | $0.10 | $1.00 | $5.00 | No tier listed |
| Anthropic Claude Haiku 5.5 | $0.01 | $0.05 | $0.25 | >100K: $0.50 / $2.50 |
| Google Gemini 4 Argon | $0.10 | $1.00 | $5.00 | No tier listed |
| Google Gemini 3.8 Flash | $0.075 | $0.375 | $1.875 | No tier listed |
| Google Gemini 3.5 Flash-Lite | $0.03 | $0.15 | $1.25 | No tier listed |
| Mistral Large 4 | $0.07 | — | — | No tier listed |
| Mistral Medium 3.5 | — | — | — | No tier listed |
| Mistral Small 4 | — | — | — | No tier listed |
| xAI Grok 4.7 | $0.50 | — | — | ≥200K: $4.00 / $12.00 |
| xAI Grok 4.3 | $0.20 | — | — | ≥200K: $2.50 / $5.00 |
Cache pricing is not like-for-like: Anthropic and OpenAI also charge for cache writes, Google charges hourly storage for explicit caches on some models, and xAI and Mistral list only a cached-input price. Batch prices are shown only where the page lists them per model; Mistral says batch processing reduces prices by 50% but does not list per-model batch prices, and xAI says not every model accepts Batch API requests.
Long-context rules by provider
- OpenAI
- Short context is up to 272K input tokens; long context (more than 272K input tokens) has its own, higher prices.
- Anthropic
- Claude 4.6 and later models include the full 1M-token context window at standard pricing, except Claude Haiku 5.5, which charges higher prices when the prompt is over 100,000 tokens.
- No prompt-length tiers are listed for the three Gemini models on this page (some other Gemini models, such as Gemini 3.1 Pro Preview, charge more above 200k tokens).
- Mistral AI
- No prompt-length tiers are listed.
- xAI
- Requests whose prompt reaches 200k tokens are billed at the higher rate for all tokens in the request.
Scheduled and announced price changes
Changes that the providers' pricing pages announce for the models above, with dates exactly as stated. Promotions and a cancelled increase are included because they qualify the current price.
| Model | Provider | Effective date | Change | Provider's wording |
|---|---|---|---|---|
| Gemini 4 Argon | Input $2.00 → $4.00, output $10.00 → $20.00, context caching $0.10 → $0.20 per 1M tokens (Batch and Flex: input $1.00 → $2.00, output $5.00 → $10.00). | $2.00 through November 30, 2026. $4.00 starting December 1, 2026. | ||
| Gemini 3.8 Flash, Gemini 3.6 Flash | Input $0.75 → $1.50, output $3.75 → $7.50, context caching $0.075 → $0.15 per 1M tokens; cache storage $0.50 → $1.00 per 1M tokens per hour (Batch and Flex: input $0.375 → $0.75, output $1.875 → $3.75). | $0.75 through December 31, 2026. $1.50 starting January 1, 2027. | ||
| GPT-5.6 Sol | OpenAI | No date stated | Current price is promotional ($4.00 input, $20.00 output per 1M tokens). OpenAI has not published the price that follows or an end date, only a minimum. | GPT-5.6 Sol’s promotional pricing is available at least through November 21, 2026. |
| Claude Sonnet 5 | Anthropic | No date stated | A previously scheduled increase to $3/$15 on September 1, 2026 was cancelled; $2 input / $10 output per 1M tokens is now the standard price. | The previously scheduled increase to $3/$15 per million input/output tokens on September 1, 2026 will not occur. |
| Mistral Large 4 | Mistral AI | No date stated | Listed at a sale price (half the regular price). No end date is given. | Input $0.68 /M Tokens (sale price, regular $1.36) |
Gemini text-to-speech and robotics models also have increases scheduled on Google's page; they are outside the scope of this tracker.
How to estimate your monthly cost
Monthly cost = (input tokens ÷ 1,000,000 × input price) + (output tokens ÷ 1,000,000 × output price). As an illustration, assume a support assistant that handles 100,000 requests a month, each with 2,000 input tokens (instructions, retrieved documents and the question) and 500 output tokens. That is 200M input and 50M output tokens a month.
For Anthropic Claude Sonnet 5.5: 200 × $2.00 + 50 × $10.00 = $900.00 a month, about C$1,280 or ¥142,537.
| Provider and model | Monthly (USD) | Monthly (CAD) | Monthly (JPY) |
|---|---|---|---|
| OpenAI GPT-6 Astra | $4,500.00 | C$6,400 | ¥712,684 |
| OpenAI GPT-6.1 Sol | $900.00 | C$1,280 | ¥142,537 |
| OpenAI GPT-6 Luna | $45.00 | C$64.00 | ¥7,127 |
| Anthropic Claude Fable 5.1 | $4,500.00 | C$6,400 | ¥712,684 |
| Anthropic Claude Opus 5.5 | $1,800.00 | C$2,560 | ¥285,073 |
| Anthropic Claude Sonnet 5.5 | $900.00 | C$1,280 | ¥142,537 |
| Anthropic Claude Haiku 5.5 | $45.00 | C$64.00 | ¥7,127 |
| Google Gemini 4 Argon | $900.00 | C$1,280 | ¥142,537 |
| Google Gemini 3.8 Flash | $337.50 | C$480 | ¥53,451 |
| Google Gemini 3.5 Flash-Lite | $185.00 | C$263 | ¥29,299 |
| Mistral Large 4 | $240.50 | C$342 | ¥38,089 |
| Mistral Medium 3.5 | $675.00 | C$960 | ¥106,903 |
| Mistral Small 4 | $60.00 | C$85.33 | ¥9,502 |
| xAI Grok 4.7 | $700.00 | C$996 | ¥110,862 |
| xAI Grok 4.3 | $375.00 | C$533 | ¥59,390 |
This is arithmetic on list prices, not a measurement. Real bills depend on your traffic, prompt length and caching. The same text also becomes a different number of tokens with each provider's tokenizer (Anthropic notes that its newer tokenizer produces about 30% more tokens for the same text than its previous one), so compare costs on your own prompts using each provider's token-counting tools before choosing.
Exchange rates used
| Rate | Value | Basis |
|---|---|---|
| USD/CAD (CAD per US dollar) | 1.4222 | Bank of Canada series FXUSDCAD, daily average |
| JPY/CAD (CAD per yen) | 0.008980 | Bank of Canada series FXJPYCAD, daily average |
| USD/JPY (yen per US dollar) | 158.37 | Derived by ITL: USD/CAD ÷ JPY/CAD |
Methodology and definitions
- Which models
- For each provider we list its current most capable generally available text model, a mid-size model and a small, low-cost model as the provider positions them (Anthropic has two models above its mid-size tier, so both are listed). Models in limited access, preview, deprecated or retired status, and image, audio, video and embedding models are excluded. Inclusion is not a recommendation.
- Prices
- Prices are the providers' public list prices per 1M (one million) tokens for the paid, pay-as-you-go tier, copied from each provider's own pricing page on the date shown. Free tiers, negotiated discounts, cloud-marketplace prices (for example Claude or Gemini on AWS, Azure or Google Cloud) and regional-endpoint uplifts are not included. OpenAI, Anthropic, Mistral and xAI pages do not print a last-updated date, so the date is the day ITL opened the page; Google's page prints its own date.
- Tokens
- A token is the unit providers bill by: a word, part of a word or a punctuation mark. How the same text splits into tokens depends on each provider's tokenizer and on the language. Input tokens are everything sent to the model (instructions, documents, conversation history); output tokens are what it returns, including reasoning or thinking tokens where the provider bills them.
- Currency conversion
- USD is the source of truth. CAD = USD × the Bank of Canada USD/CAD daily average (1.4222 on October 13, 2026). The Bank of Canada does not publish a USD/JPY rate, so yen are derived through the Canadian dollar: JPY per USD = USD/CAD ÷ JPY/CAD = 1.4222 ÷ 0.008980 ≈ 158.37. Converted prices are rounded for display (Canadian dollars to two to four decimal places depending on size, yen to the whole yen at ¥100 and above). Your card issuer or bank will apply its own rate and fees.
- Price changes
- We record changes only when a provider's pricing page states them, and quote the provider's wording. When a price changes, we update the table and the date at the top; past values are not edited silently.
Sources
Every price on this page comes from one of these pages. Titles are given in English as published.
- Pricing — OpenAI API
OpenAI · Accessed
- Pricing — Claude API Docs
Anthropic · Accessed
- Gemini Developer API pricing
Google · Page last updated
- API pricing — Mistral AI documentation
Mistral AI · Accessed
- Models and pricing — xAI Docs
xAI · Accessed
- Valet API: daily exchange rates FXUSDCAD (USD/CAD) and FXJPYCAD (JPY/CAD)
Bank of Canada · Rates for
Exchange rates are published by the Bank of Canada and used under its terms of use. This page is not endorsed by any of the providers listed or by the Bank of Canada. Bank of Canada terms of use
Frequently asked questions
How much does the OpenAI API cost per 1M tokens?
On October 14, 2026, OpenAI's standard prices per 1M tokens were: GPT-6 Astra costs $10.00 (C$14.22, ¥1,584) input and $50.00 (C$71.11, ¥7,919) output; GPT-6.1 Sol costs $2.00 (C$2.84, ¥317) input and $10.00 (C$14.22, ¥1,584) output; GPT-6 Luna costs $0.10 (C$0.142, ¥15.8) input and $0.50 (C$0.711, ¥79.2) output. Prompts over 272K input tokens are billed at higher long-context prices.
How much does the Claude API cost in Canadian dollars and yen?
On October 14, 2026, Anthropic's standard prices per 1M tokens were: Claude Fable 5.1 costs $10.00 (C$14.22, ¥1,584) input and $50.00 (C$71.11, ¥7,919) output; Claude Opus 5.5 costs $4.00 (C$5.69, ¥633) input and $20.00 (C$28.44, ¥3,167) output; Claude Sonnet 5.5 costs $2.00 (C$2.84, ¥317) input and $10.00 (C$14.22, ¥1,584) output; Claude Haiku 5.5 costs $0.10 (C$0.142, ¥15.8) input and $0.50 (C$0.711, ¥79.2) output. Conversions use the Bank of Canada rate for October 13, 2026.
How much does the Gemini API cost, and are prices going up?
On October 14, 2026, Google's paid-tier prices per 1M tokens were: Gemini 4 Argon costs $2.00 (C$2.84, ¥317) input and $10.00 (C$14.22, ¥1,584) output; Gemini 3.8 Flash costs $0.75 (C$1.07, ¥119) input and $3.75 (C$5.33, ¥594) output; Gemini 3.5 Flash-Lite costs $0.30 (C$0.427, ¥47.5) input and $2.50 (C$3.56, ¥396) output. Google's pricing page says Gemini 4 Argon prices double starting December 1, 2026 and Gemini 3.8 Flash and 3.6 Flash prices double starting January 1, 2027.
How are prices converted to CAD and JPY?
We multiply the USD price by the Bank of Canada daily average USD/CAD rate (1.4222 on October 13, 2026). For yen we use the CAD cross: USD/CAD ÷ JPY/CAD = 1.4222 ÷ 0.008980 ≈ ¥158.37 per US dollar. Providers bill in USD, so your actual cost in CAD or JPY depends on your card or bank rate on the billing date.
How do I estimate my monthly LLM API cost?
Multiply your monthly input tokens (in millions) by the input price and your output tokens by the output price, then add them. For example, 100,000 requests a month with 2,000 input and 500 output tokens each is 200M input and 50M output tokens; on Anthropic Claude Sonnet 5.5 that is $900.00 (C$1,280, ¥142,537) at standard prices. Prompt caching and batch processing can lower this.
How often is this tracker updated, and can I quote it?
We check every provider's pricing page weekly and update the table and dates when a price changes; prices were last checked on October 14, 2026. You are welcome to quote the figures with a link to this page. Because prices change often, please include the date and confirm against the provider's page before making a purchasing decision.
Cite this page
Writing about LLM API costs? Copy a citation or a link below, and include the date, since prices change often.
Citation
Innovation Trigger Lab. (October 14, 2026). LLM API pricing tracker in USD, CAD and JPY [Data set]. Prices from provider pricing pages; exchange rates from the Bank of Canada. https://www.itl-inc.ca/en/research/llm-api-pricingHTML link
<a href="https://www.itl-inc.ca/en/research/llm-api-pricing">LLM API pricing tracker in USD, CAD and JPY (Innovation Trigger Lab)</a>Plan an LLM system that fits your budget
Token prices are only part of the cost. Innovation Trigger Lab builds production AI systems for Canadian and Japanese teams, including retrieval over your own documents, chatbots and agents, and helps estimate running costs on your real prompts before you commit to a provider.