Free calculator · runs in your browser
Subscription or pay-per-use API?
A chat plan is a flat fee; the API charges per token. Enter how you use it and see which is cheaper, and how many messages a day it takes to break even.
Per month
API is cheaper
| Cost per message (API) | $0.012 |
| API at 30 messages/day | $10.94 |
| Claude Pro | $20.00 |
| Break-even | 55 msgs/day |
| Cheaper for you | API |
Sources
- platform.claude.com/docs/en/about-claude/pricing
- developers.openai.com/api/docs/pricing
- ai.google.dev/gemini-api/docs/pricing
- claude.com/pricing
- one.google.com/about/plans
- api.vultr.com/v2/plans
- hetzner.com/cloud/regular-performance/
- digitalocean.com/pricing/droplets
- digitalocean.com/pricing/app-platform
- digitalocean.com/pricing/managed-databases
- railway.com/pricing
- brevo.com/pricing/
The same usage on every model
| Model | API per month | Break-even msgs/day |
|---|---|---|
| Claude Haiku 5.5 Anthropic | $0.547 | 1,096 |
| GPT-6 Luna OpenAI | $0.547 | 1,096 |
| Gemini 3.1 Flash-Lite Google | $1.50 | 399 |
| Gemini 3.5 Flash-Lite Google | $2.19 | 274 |
| Gemini 3.8 Flash Google | $4.10 | 146 |
| Claude Sonnet 5.5 Anthropic | $10.94 | 55 |
| GPT-6.1 Sol OpenAI | $10.94 | 55 |
| Gemini 3.1 Pro (Preview) Google | $12.04 | 50 |
| Claude Opus 5.5 Anthropic | $21.89 | 27 |
| Claude Fable 5.1 Anthropic | $54.72 | 11 |
| GPT-6 Astra OpenAI | $54.72 | 11 |
How it works
Cost per message = input tokens × input price + output tokens × output price (per million tokens), with the cached share of input at the cache-read price. The monthly API cost is that × messages per day × 30.4. Break-even is the number of messages a day at which the API costs the same as the plan.
Chat plans include things the API doesn't (apps, file handling, web search, higher limits on heavier tiers) and the API includes things plans don't (no usage caps, automation). This compares cost only.
Questions
Why do long chats cost so much more on the API?
Every message re-sends the conversation so far as input. A 20-message chat sends the first message 20 times. That is why the input-tokens default here (3,000) is much larger than a typical question. Use a higher number for long chats or attached documents.
Is the subscription really unlimited?
No. Chat plans have usage limits that reset over hours or days, and heavier tiers raise those limits. If you hit the limits often, compare against the higher tier or the API.
Where is ChatGPT Plus?
We only list prices we could read on the provider's own pricing page, and OpenAI's consumer pricing pages didn't load for us when we checked. Pick 'Another plan' and enter the price you pay.