FAQ
Frequently asked questions
Answers about Tokener CA: the API and its three protocols, API keys, pricing and billing, and how your data is handled.
About
What is Tokener CA?
Tokener CA is an API for open-weight AI models. Create an API key and call the models from your own code or from the tools you already use, then pay the way that suits you: as you go, with funds you add to your balance, or with a monthly Agent plan that includes usage up to a 5-hour and a weekly limit.
Who runs Tokener CA?
Tokener CA is operated by ShowKey Information Systems Inc., a Canadian company based in the Greater Toronto Area, Ontario.
Which models can I use?
A short, curated catalog of open models. The models page lists each one with its price, context window and capabilities.
What do BF16 and FP8 mean on a model card?
BF16 means 16-bit weights that are not quantized: the model as its maker released it. FP8 means 8-bit weights, often the version the model's maker publishes; quality stays close to 16-bit, and it runs faster. Precision is how many bits store each of the model's weights. Fewer bits make a model cheaper to serve but can lower answer quality, especially in coding, tool use and long inputs. We show it so you know what you get. It applies to requests under the default Data handling setting; an account that allows all suppliers may sometimes be served at a different precision.
Using the API
Does it work with the OpenAI SDK?
Yes. Point the OpenAI SDK at https://a.tokener.ca/v1, use a Tokener API key and a model id from the models page. Both OpenAI-compatible (Chat Completions) and OpenAI Responses are supported.
Can I use the Anthropic SDK?
Yes. The API also speaks Anthropic-compatible (Messages). Set the SDK's base URL to https://a.tokener.ca and use the same API key.
Do I need a different key for each protocol?
No. One key works on all three protocols, which share the same balance, prices and usage page.
How do I get an API key?
Sign up, verify your email, then create a key on the API keys page. The full key is shown once, when you create it, so copy it then.
I lost my API key. Can I see it again?
No. Tokener stores only a hash of the key, so it cannot be shown again. Revoke the lost key and create a new one; revoking takes effect immediately.
Why is a model marked Temporarily unavailable?
Our automatic checks cannot reach that model right now. It is shown again as soon as a check succeeds, usually within minutes. The API still accepts requests for it, and a request that fails is not billed; in the Playground it cannot be selected until it is back. Pick another model or retry later.
How do you measure uptime?
The uptime on the Models pages is the share of our automatic test requests in the last 3 UTC days (today so far and the two days before) that got a complete answer. We send them about every 30 minutes from our server to the suppliers your requests use by default; supplier timeouts and rate limits count as failures. A short outage between two checks can be missed. The card always names the period it covers: usually the last 3 days, with a block for each day; some models may show a shorter period (the last 24, 12 or 6 hours) or the last 7 days, with the same definition for that period.
Why does the Playground ask before using some models?
By default your requests go only to suppliers that commit to not storing request content and not using it for training. When a model is temporarily available only from suppliers without that commitment, the Playground asks before using them; Continue allows them for that model for the rest of the session, billed at the current price. An API request for such a model that fails answers model_limited_by_data_handling and is not billed. To allow those suppliers for every request, change Settings → Data handling.
Can I try a model without writing code?
Yes. The Playground in your account lets you chat with any model in the catalog, attach files and copy the equivalent code.
Pricing and billing
How does pricing work?
You pay per token. Each model has a price per 1M input tokens and per 1M output tokens, and a lower price for cached input tokens where one is listed.
Is there a subscription or a monthly fee?
Pay as you go has no subscription and no monthly fee. If you prefer a fixed monthly price, the Agent plan covers your usage up to a 5-hour and a weekly limit; when a limit is reached your request continues on your balance if you allow it, otherwise it waits for the next reset.
What is the Agent plan?
A monthly subscription that pays for your API and Playground requests up to a 5-hour and a weekly usage limit, on the same API key and the same endpoints. Usage is counted at the prices you would pay from your balance. You choose a tier on the pricing page, manage the plan on the Plan page of your account, and can cancel at any time; the plan then ends at the end of the period you paid for.
Is the Agent plan only for coding tools?
No. It covers any API call: coding agents, monitoring loops, scheduled pipelines and your own apps all count, on the same key and every model.
Can I change my Agent plan tier?
Yes, on the Plan page of your account. An upgrade to a tier with a higher monthly price starts at once: you pay only the prorated difference for the rest of the current billing period, and the new limits apply immediately. A switch to a tier with a lower monthly price takes effect at the end of the current period, and you keep your current tier until then.
Can I get a refund for the Agent plan?
Agent plan payments are non-refundable, except where required by law. If you cancel, your plan stays active until the end of the period you paid for, and you are not charged again. Unused balance top-ups follow the separate refund rule on the Refund page.
What happens when my plan reaches a limit?
If "At a limit, continue on my balance" is on, your requests continue on your balance. If it is off, the API answers with a 402 error and a Retry-After header that says how many seconds remain until the next reset. The Plan page shows both windows and when each one resets.
Can one key stay on the balance while my agents use the plan?
Yes. On the API keys page, set that key's Funding to Balance only. It is then always billed per token from your balance and never uses the plan; your other keys keep using the plan first.
Which currencies can I pay in?
Your balance is held in US dollars. You can pay in USD or CAD; a CAD payment is converted to a USD credit at a rate locked at checkout. Payments are processed by Stripe.
Am I charged for failed requests?
No. A request that fails is not charged.
Are reasoning tokens charged?
Yes. When a model reasons before answering, those tokens count as output tokens. On models that offer reasoning controls you can lower the effort or turn it off.
Where can I see what I spent?
The Usage page lists each request with its model, protocol, token counts and cost. The Transactions page shows every change to your balance.
Do new accounts get free credit?
A welcome credit may be offered to new accounts that verify their email. Whether it is on, and its amount, can change; after you sign up, your account shows your own status.
Data
Do you store my prompts or use them for training?
Tokener does not store the prompts you send to the API or the outputs the models return, and does not use them for training. We record what billing needs: the model, token counts, timing and a request id. The support chat is different — those conversations are recorded, as the Privacy Policy explains.
Is the support chat recorded?
Yes. Support conversations are recorded so that staff can answer you and follow up, as the Privacy Policy explains. Attachment contents are not stored.