Starter
Free2M tokens a week
Start free- A couple of coding tasks a week, plus daily chat
- A one-time 5M token welcome grant for Code, around a full day of work
- At the limit, usage stops until the week resets
Each plan is a weekly pool of tokens that Chat and Code share. Start free, and move up when you need more each week.
Prices in Australian dollars, including GST
Free2M tokens a week
Start free$25a month8M tokens a week
Choose Standard$125a month48M tokens a week
Choose SuperFor organisations · priced to your contract
Your whole team on one contract and one invoice, on our hardware in Melbourne or in a dedicated instance.
For organisations · scoped
We train an open-weights model on your material and run it in your own instance, and the price depends on the work and where it runs.
In every plan
For developers · pay per token
You pay per million tokens, and a cached prompt costs a twentieth of fresh input to read again. You manage keys and daily usage in Matilda Settings. For dedicated capacity, contact sales.
| Input | $1 |
|---|---|
| Output | $5 |
| Prompt cache write, 5 minutes | $1.25 |
| Prompt cache write, 1 hour | $2 |
| Prompt cache hit or refresh | $0.05 |
Questions
Yes. Starter is free, with 2M tokens a week for Chat and Code and a one-time 5M token welcome grant for Code. Move to Standard or Super when you need more each week.
Chat and Code draw on the same pool of tokens, and it resets each week. A coding task reads and writes far more than a chat message, so on Standard 8M tokens is around 9 coding tasks or around 2,800 chat messages.
The size of the pool and what happens at the limit. Standard costs $25 a month for 8M tokens a week, and usage stops at the limit until the week resets. Super costs $125 a month for 48M tokens a week, and you can switch on pay-as-you-go past the limit, with a spending cap you set.
Plan prices include GST. API prices exclude it. Every price is in Australian dollars.
Per million tokens on matilda-chat: $1 for input and $5 for output. Prompt caching costs $1.25 to write for 5 minutes, $2 to write for an hour and $0.05 for each hit or refresh, all excluding GST.
Through a contract with us that covers your team’s seats, invoicing and support, on our hardware in Melbourne or in a dedicated instance. Contact sales to scope it.
Matilda runs on hardware we own in Melbourne, which serves Chat, Code, Desktop and the API, and a model of your own runs in our facility or in your data centre.
No. Our privacy policy commits us to never using the contents of your prompts or files to train shared models, and to never selling your data. See trust and security for how we hold it.
We price it by the work, where it runs and how much of the stack you take, from Matilda under your brand to hardware built to our design, and we set the price when we scope the project. See the deployment options, or contact sales.