Each plan includes a different amount of usage. Here's how usage is measured and what happens when you reach your limit.
What each plan includes
- Free — everyday chat with Use AI Instant at limited capacity; the limit is dynamic and not disclosed, and the service is provided as-is. Free doesn't include premium model usage.
- Pro — a monthly allowance of premium model usage, plus unlimited messaging with Use AI Instant, subject to fair use.
- Max — everything in Pro, with 5x the premium model usage.
For the models each plan unlocks, see the pricing page. Models outside your plan are marked in the model picker with the plan that unlocks them; selecting one shows an upgrade option — your message isn't sent and you aren't charged.
How your allowance is applied
Your monthly allowance is applied weekly — a share each week, so one busy stretch never drains the whole month.
During the 7-day trial, the limit applies daily, proportional to the monthly allowance — see How the trial works. After the trial converts, the weekly allowance takes over.
What counts toward your usage
Usage is measured by compute, so not every message costs the same:
- Model — more capable models use more per reply.
- Conversation length — long chats re-send their history with every message, so late replies use more. A new chat per topic makes your allowance go further.
- Files and tools — uploaded files, web research, and other heavy tools add to what a request uses.
Checking where you stand
Usage shows as a percentage in your profile, and in the model picker as you approach the limit. 100% means your allowance is used up for the week — or the day, on trial.
How many messages does a plan buy?
Estimates for a typical message, model by model — and where unlimited kicks in.
|
Model |
Pro |
Max |
|---|---|---|
|
Use AI Instant — never counts toward your usage |
Unlimited |
Unlimited |
|
Claude Fable 5.1 |
Max only |
≈ 1,904 / mo |
|
Claude Opus 5 |
≈ 761 / mo |
≈ 3,809 / mo |
|
Claude Opus 4.8 |
≈ 761 / mo |
≈ 3,809 / mo |
|
Claude Sonnet 5 |
≈ 1,904 / mo |
≈ 9,523 / mo |
|
GPT-6 Astra |
≈ 380 / mo |
≈ 1,904 / mo |
|
GPT-5.6 Sol |
≈ 952 / mo |
≈ 4,761 / mo |
|
GPT-5.5 |
≈ 666 / mo |
≈ 3,333 / mo |
|
GPT-5.4 |
≈ 1,333 / mo |
≈ 6,666 / mo |
|
Gemini 3.8 Flash |
≈ 5,079 / mo |
≈ 25,396 / mo |
|
DeepSeek V4 Pro |
≈ 4,040 / mo |
≈ 20,202 / mo |
|
Kimi K3 |
≈ 1,269 / mo |
≈ 6,349 / mo |
|
Kimi K2.6 |
≈ 4,519 / mo |
≈ 22,598 / mo |
|
Grok 4.6 |
≈ 2,666 / mo |
≈ 13,333 / mo |
|
GLM 5.2 |
≈ 3,703 / mo |
≈ 18,518 / mo |
Image generation
Approximate image generations at standard quality (~1024px), matching the pricing page calculator:
|
Model |
Pro |
Max |
|---|---|---|
|
Nano Banana 2 |
≈ 298 / mo |
≈ 1,492 / mo |
|
Nano Banana 2 Lite |
≈ 588 / mo |
≈ 2,941 / mo |
|
GPT Image 2 |
≈ 377 / mo |
≈ 1,886 / mo |
|
FLUX.2 Pro |
≈ 666 / mo |
≈ 3,333 / mo |
|
MAI-Image-2.5 |
≈ 400 / mo |
≈ 2,000 / mo |
|
Krea 2 Medium Turbo |
≈ 1,333 / mo |
≈ 6,666 / mo |
|
Krea 2 Medium |
≈ 666 / mo |
≈ 3,333 / mo |
Estimates only — actual counts vary with conversation length, attachments, reasoning effort and web search.
What happens at the limit
If you reach your limit mid-conversation, the reply in progress completes on your premium model; then you're switched to Use AI Instant, with a notice in the chat, and can keep chatting right away.
From there you can:
- Keep chatting with Use AI Instant — unlimited on Pro and Max, subject to fair use; see Use AI Instant and fair use.
- Wait for the reset — your allowance returns automatically each week (daily on trial).
- Upgrade to Max — 5x the premium usage of Pro.
- Add a Usage Boost (coming soon, Pro and Max) — extra allowance on top of your plan; see Usage Boost.
Unlimited Instant usage on paid plans is subject to our fair usage policy.
After the reset
Premium models re-enable automatically. You stay on Use AI Instant until you pick your premium model again in the model picker.