Napsix's first model
Mato V1.0:
reasons
Napsix's own model. Built on Qwen3.8 27B, served on our own NVIDIA H100 infrastructure. Fast, native tool-calling, and free on every plan: use it in XIA Chat or via API at napsix/mato-v1.0.
No setup: it's already in XIA Chat, in agents and in the API.
How Mato works
Illustrative tour with sample data; not a live run.
Native tools, not prompts
Mato was trained with Napsix tools inside: it calls them on its own, in the right order and on your data.
Sheets
Reads and writes sheets: compares, calculates and leaves the table ready.
CRM
Searches, loads and updates contacts and opportunities.
Mail
Reads threads, summarizes and drafts replies in your tone.
Crawl
Searches the web and reads pages or PDFs when it lacks context.
Docs
Writes documents and reports from what it found.
Forms
Creates and fills forms; answers land in your CRM.
It runs your apps, not just mentions them
The same catalog your agents already use. Mato decides which app to touch, reads or writes it, and hands you the result.
Is the annual plan still available?
XIAReply with the current policy
Illustrative sequence; connections are your own.
Generic model vs Mato on Napsix tasks
| Criterion | Generic model | Mato V1.0 |
|---|---|---|
| Access to your data | Needs connectors you build | Napsix tools natively |
| Tool calls | Manual or nonexistent | Verified on Forms, Sheets and CRM |
| Context per task | One model for everything | Routed to the right size |
| Cost | Paid per token | $0 per token, every plan |
Illustrative workflow comparison; it does not attribute unverified limits to other models.
Spec sheet
What makes Mato different from any generic model.
| Specification | Mato V1.0 |
|---|---|
| Context | 256k native · up to 1M with rope scaling |
| Input | Text · vision (image and video): being validated |
| Tool-calling | Verified on Napsix tools |
| Thinking | Flexible, on by default in direct mode |
| Price | $0 per token, on every plan |
| Served on | NVIDIA H100 80GB, our own infrastructure |
Data verified as of 2026-09; multimodal vision is being validated.
- ~2s
- Short answers
- ~30 tok/s
- Streaming
- 580k
- KV cache tokens
- 17.7×
- Concurrency
Measured in production (2026-09) on Daytona H100. Varies with load.
Who we built it with
The base model family and the infrastructure: each piece with the people who do it best.
Alibaba Cloud
Their Qwen model family is the base Mato V1.0 is trained on.
Daytona · Startup Grid
NVIDIA H100 infrastructure through the Startup Grid Program: the lift that makes serving it for free possible.
Base: Qwen 3.8 27B · Apache-2.0 · official FP8 quantization.
It works with your apps
Managed integrations from your account (OAuth or API key), on top of the native tools.
- Gmail
- Slack
- Notion
- HubSpot
- Google Sheets
- Google Drive
- Stripe
- Shopify
- Tienda Nube
- Trello
- Airtable
Napsix apps
I have prepared the next steps.
Frequently asked questions
Is Mato really free?
What if it's not available?
Does it read images or video?
Which tools does it run?
Do I need to set anything up?
Can I use it in production?
Your next step
From evaluation to a concrete integration.
Start with an account or talk to us about the scope of your team and application.
