The short answer
Whizi has three plans. All of them include web, iOS and Android access on one login, and all of them include file upload. The difference is which models you can reach and how much you can send.
| Starter | Pro | Powerhouse | |
|---|---|---|---|
| Monthly | $15.99 | $29.99 | $49.99 |
| Yearly, per month | $10.99 | $19.99 | $34.99 |
| Yearly, billed once | $131.88 | $239.88 | $419.88 |
| Credits per month | 600 | 2,000 | 8,000 |
| Text models | 3 | about 40 | 280 plus |
| Image generations | none | 100 | 500 |
| AI voice | 10 minutes | 80 minutes | 500 minutes |
| AI video | none | none | included |
| AI music | none | none | included |
| Side-by-side chats | no | no | yes |
| Agent personas | no | yes | yes |
For context, a single ChatGPT Plus subscription is $20 per month and a single Claude Pro subscription is $20 per month. Whizi Pro at $19.99 billed yearly covers both model families plus Gemini, Grok, DeepSeek, Llama, Mistral, Kimi and Qwen.
The trial
New accounts start with a 7 day trial for $0.99. It is a paid trial rather than a free one, which filters out most of the abuse that forces other products into aggressive rate limits.
The trial converts to the plan you selected at signup unless you cancel before it ends. Cancelling during the trial ends the subscription on the spot, with nothing further charged; it is the one cancellation that does not wait for the end of a term.
You can cancel from account settings on the web at any time. Subscriptions bought through the iOS or Android app are cancelled through Apple or Google instead, because the store owns that billing relationship. See billing and cancellation.
What counts against a limit and what does not
Three separate meters exist on an account, and they do not share a pool.
| Meter | What spends it | Reset |
|---|---|---|
| Credits | One per text message, multiplied by the model rate | Monthly, when the UTC calendar month rolls over |
| Image generations | One per generated image | Monthly |
| Voice minutes | Time spent in voice mode | Monthly |
Video and music generation are Powerhouse features rather than separate meters on every plan; a standalone video or music job also spends one credit.
Things that spend nothing: uploading a file, reading or re-reading a past conversation, sharing a conversation, switching models mid-thread, and editing a prompt before you send it. Web search has no separate charge of its own; the search-grounded models such as Sonar Pro price their searches into their per-message credit rate.
One thing that costs double: side-by-side comparison. Two models each produce an answer, and each is billed at its own credit rate, so comparing Claude Sonnet 5 against GPT-5.6 Sol costs 30 credits rather than 10 or 20.
Which plan actually fits
The honest version, given the model tiers rather than the marketing.
Starter makes sense if you want a cheap, fast assistant and the ability to ask questions about your own documents, and you do not care which model answers. Three models, all at 1 credit, 600 messages per month. It does not include any Anthropic model, so if Claude is the reason you are here, this is not the plan.
Pro is the plan most people want. It is the tier that adds Claude Sonnet 5 and the GPT-5.6 Terra class, which together cover the work that a ChatGPT Plus and Claude Pro pair would cover, at less than either one alone when billed yearly. 2,000 credits is a serious allowance: 200 Claude Sonnet messages plus effectively unlimited fast-tier use.
Powerhouse is for people who specifically need the frontier tier, the media generation, or side-by-side comparison. Claude Opus 5, GPT-6 Astra, AI video and AI music all live here and nowhere else. If you are not going to use any of those four things, Pro is the better value and the model list is not the reason to upgrade.
- Yearly billing is roughly a 30 percent discount across all three plans
- The trial is 7 days for $0.99 and converts unless cancelled
- Starter includes no Anthropic model
- Pro is the tier that adds Claude Sonnet and GPT-5.6 Terra
- Claude Opus, GPT-6 Astra, video, music and side-by-side are Powerhouse only
- Credits, image generations and voice minutes are three separate meters
- Uploading files costs nothing; web search is priced into the search models' rate
Frequently asked questions
How much does Whizi cost?
Starter is $15.99 per month or $10.99 per month billed yearly, Pro is $29.99 or $19.99 billed yearly, and Powerhouse is $49.99 or $34.99 billed yearly. Every new account starts with a 7 day trial for $0.99. Yearly billing charges once at $131.88, $239.88 or $419.88 respectively.
Is Whizi cheaper than paying for ChatGPT Plus and Claude Pro separately?
Yes, substantially. ChatGPT Plus and Claude Pro are $20 per month each, so $40 per month for two model families. Whizi Pro at $19.99 per month billed yearly covers both plus Gemini, Grok, DeepSeek, Llama, Mistral, Kimi and Qwen, which is roughly half the price for far more models. The trade is the credit allowance: unlimited use of one model is not the same shape of deal as a metered allowance across forty.
Can I change plans mid-month?
Yes. Upgrading takes effect immediately so the new models and the larger allowance are available right away. Downgrading takes effect at the end of the current paid term, so you keep what you paid for until the period you already bought runs out.
What happens to my conversations if I downgrade?
They stay. Downgrading changes which models you can send new messages to, not what you can read, so past conversations with a model you no longer have access to remain fully readable. You just cannot continue them on that model until you upgrade again, and the app will offer a model you do have access to instead.
Is there a free plan?
Yes, since 2026-09-02, and it does not run out for good. A new account gets 7 free messages on its first day, then 3 more at every UTC midnight, spendable on Whizi AI or Auto rather than on the whole catalogue, and a free turn is capped at about 5,000 tokens in and 5,000 out. The way past those limits is the 7 day trial for $0.99. A paid trial is also a much stronger filter against automated abuse than an open free tier, which is what keeps the aggressive rate limiting other free tiers force on everybody else off this one.