The short answer
Whizi includes more than 280 AI models from OpenAI, Anthropic, Google, xAI, DeepSeek, Meta, Mistral, Moonshot, Qwen, Z.ai and others under a single subscription. Access is tiered across three plans.
| Plan | Price per month | Text models | Monthly credits |
|---|---|---|---|
| Starter | $15.99, or $10.99 billed yearly | Whizi AI, GPT-5.6 Luna, Gemini 3 Flash | 400 |
| Pro | $29.99, or $19.99 billed yearly | Everything in Starter plus Claude Sonnet 5, GPT-5.6 Terra, Grok 4.6, Kimi K3 and roughly 36 more | 2,000 |
| Powerhouse | $49.99, or $34.99 billed yearly | The full catalogue, including Claude Opus 5 and GPT-5.6 Sol | 8,000 |
Every plan covers web, iOS and Android from one login. New accounts start with a 7 day trial for $1.
What Starter includes
Starter is deliberately narrow: three text models chosen because they are fast, cheap to run, and cost 1 credit per message, so a 400 credit allowance goes a long way.
| Model | Provider | Credits per message |
|---|---|---|
| Whizi AI | Whizi house assistant | 1 |
| GPT-5.6 Luna | OpenAI | 1 |
| Gemini 3 Flash | 1 |
Whizi AI is the default assistant. It runs on GPT-5.6 Luna with a Whizi system prompt, and the raw model is offered separately in the full model list for anyone who would rather have no persona layer on top.
Starter also covers image, PDF and Word document upload, which is the feature most people actually came for. What it does not include is image generation, side-by-side comparison, or any frontier model.
What Pro adds
Pro is a curated tier rather than "everything cheap". The rule is one or two current flagships per model family, each fast tier, and the high volume workhorses. It works out to roughly 40 text models.
OpenAI. GPT-5.6 Terra, GPT-5.2, GPT-5.5, GPT-5 Mini, GPT-4.1, GPT-4.1 Mini, GPT-4o, GPT-4o Mini, GPT-OSS-120B.
Anthropic. Claude Sonnet 5, Claude Sonnet 4.6, Claude Sonnet 4.5, Claude Haiku 4.5. Sonnet is the Pro ceiling, and the Opus class is Powerhouse.
Google. Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.1 Pro Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash Lite.
Everyone else. Grok 4.6 from xAI, DeepSeek V3.2, DeepSeek Chat V3.1, DeepSeek R1, Llama 4 Maverick, Llama 4 Scout, Llama 3.3 70B, Mistral Large 2512, Mistral Medium 3.1, Mistral Small 3.2, Mistral Nemo, Qwen3.7 Plus, Qwen3 235B, Kimi K3, Kimi K2.5, GLM-5, and Sonar Pro.
Pro also unlocks the agent personas, which are a feature rather than a model tier: the Canadian CPA agent, the study mentor, the writing coach and the math tutor all run on the same base model, and what you are buying is the prompt, the token budget and the reasoning depth. Pro adds 100 image generations per month on top.
What Powerhouse unlocks
Powerhouse opens the remainder of the catalogue. Anything not named in the Starter or Pro lists above resolves to Powerhouse, which is roughly 240 further models. The categories held back on purpose:
- The Opus class. Claude Opus 5, and the 4.x Opus generation. 20 credits per message.
- GPT-5.6 Sol. Held back deliberately as the trophy next to Claude Opus, at 20 credits per message. Terra takes the recommended row on Pro, because the top of the picker should never be locked.
- The frontier reasoning tier. Grok 4.5, Qwen3 Max Thinking, GLM-5.1, Kimi K2 Thinking, Intellect-3.
- Coding specialists. GPT-5.2 Codex, Kimi K2.7 Code, and the Qwen3 Coder and Devstral families.
- The 400B class open weights, and the routing models.
Powerhouse is also where the non-text generation lives: side-by-side model chats, AI music generation, 500 image generations, 500 minutes of AI voice, and 100 AI video generations per month.
What a model costs in credits
Every plan spends from a single monthly credit allowance, and the cost of a message depends on which model answered. Most of the catalogue sits at 1 or 2 credits, and the scale runs up to 80 for the oldest Opus generation. The multiplier tracks what the model actually costs to run, so the cheap workhorses stay effectively unlimited and the frontier models are metered.
| Model | Credits per message |
|---|---|
| Whizi AI, GPT-5.6 Luna, Gemini 3 Flash, DeepSeek V3.2, Qwen3.7 Plus | 1 |
| GLM-5 | 2 |
| GPT-5.6 Terra, Claude Haiku 4.5 | 4 |
| Grok 4.6 | 6 |
| Claude Sonnet 5, GPT-5.2, Gemini 3.1 Pro, Kimi K3 | 10 |
| Claude Opus 5, GPT-5.6 Sol, Sonar Pro | 20 |
The practical consequence: 2,000 Pro credits is 200 Claude Sonnet 5 messages, or 500 GPT-5.6 Terra messages, or 2,000 messages on the 1 credit tier, or any mix of the three. 8,000 Powerhouse credits is 400 Claude Opus 5 messages.
Image, video and music generation are not credit charged. They draw on their own monthly counts, listed per plan above. The full per model breakdown is in the credits reference.
For a single model, the model reference carries a page each: the list rate per million tokens, the cost of one standard answer, the context window, the credit cost, and the plan that unlocks it. For example Claude Opus 5, Claude Sonnet 5 and GPT-5.6 Terra.
How access is enforced
Model access is decided on the server, not in the interface. The worker ships each model minimum tier on the models endpoint, and the same function backs the gate on the chat route, so a locked model cannot be reached by editing a request.
The fallback for an unrecognised model id is Powerhouse rather than Pro. That matters when the catalogue gains a model: a new arrival is Powerhouse-only until it is explicitly promoted, which is the safe direction. It also means the Starter and Pro lists above are exhaustive rather than indicative.
The web app and the mobile apps read the same catalogue, so a model available on the iPhone app is available on the web on the same plan.
- Starter covers three 1 credit models plus file upload
- Pro is the tier that adds Claude Sonnet and the GPT-5.6 Terra class
- Claude Opus and GPT-5.6 Sol require Powerhouse
- Credits are one shared allowance, and most models spend 1 or 2 per message
- Image, video and music generation are counted separately, not in credits
- Anything not named in the Starter or Pro list requires Powerhouse
Frequently asked questions
Does Whizi include both ChatGPT and Claude in one subscription?
Yes. Pro and Powerhouse both include OpenAI and Anthropic models in the same workspace, so you can send a prompt to GPT and then switch the same conversation to Claude without losing context. Pro reaches Claude Sonnet 5 and GPT-5.6 Terra, and Powerhouse reaches Claude Opus 5 and GPT-5.6 Sol. Starter includes OpenAI and Google models but no Anthropic model.
How many AI models does Whizi have?
The catalogue is over 280 text models plus the image, video, music and voice generators. Three are on Starter, roughly 40 on Pro, and the remainder on Powerhouse. The number moves as providers add and retire models, and the model picker inside the app is always the live list.
Is Claude Opus available on the Pro plan?
No. Sonnet is the Pro ceiling for Anthropic models, and the entire Opus class requires Powerhouse. This is a deliberate split rather than a technical limit: Powerhouse needs a frontier model to sell, and Opus at 20 credits per message costs several times what Sonnet costs to run.
What happens when I run out of credits?
Sending stops until the allowance resets at the start of your next billing month, and the app offers an upgrade at that point. Because roughly half the catalogue costs 1 credit while the frontier tier costs 20, the cheapest way to stretch an allowance is to route routine work to the 1 credit models and save the expensive ones for the messages that actually need them.
Do the mobile apps have the same models as the web app?
Yes. The web app, the iOS app and the Android app read the same server-side catalogue and the same tier gate, so a model available on one is available on all three at the same plan level. Your subscription and your credit allowance are shared across every device on one login.