Ordalin
Submit a tool

Reviewed tool profile

Agentaus 1

Agentaus 1 is Trellis Data's 220B-parameter Australian AI model, running in a document-grounded assistant with data kept onshore.

Sources checked

Visit website ↗
Pricing
Prices are in Australian dollars and exclude GST. Free costs A$0 and gives a daily allowance of protected prompts plus a monthly top-up of Claude and ChatGPT credit. Standard costs A$24 a month for 5× the free prompts, and Professional costs A$107 a month for unlimited prompts and private agents. Team plans cost A$24 per seat for 2–19 seats, and Business costs A$107 per seat for 2–999 seats. Enterprise and on-premises deployments are custom priced. The API costs A$1.70 per million input tokens and A$6.70 per million output tokens. Official source 1 ↗
Available as
Agentaus is a web app hosted in Australian data centres. Businesses can also install it on their own systems or run it fully offline on high-security networks. · The Agentaus API is compatible with OpenAI's chat completions API and is billed per token. It currently serves one model, with the id agentaus.v1. · Inside the app, MCP connectors let Agentaus use information from your other tools. OpenCode can use the Agentaus API as its model provider. Official source 1 ↗ Official source 1 ↗ · Official source 2 ↗ Official source 1 ↗ · Official source 2 ↗
Primary group
Research & Data

Also consider

See all 6 similar tools ↓

What it does

Agentaus 1 is a 220-billion-parameter AI model from Trellis Data, an AI company based in Canberra. It was announced on 10 November 2026 and runs inside Agentaus, Trellis Data's AI assistant. Trellis Data says the model is about 50% larger than its previous one and aims for better accuracy, deeper multi-step reasoning and longer outputs, while still running on smaller compute clusters. In Agentaus you can ask questions, search and summarise your documents, and draft work. Everything is processed in Australian data centres on Trellis Data's own cloud, and your data is never used for training. It is offered on free, subscription, enterprise and API tiers.

Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗

Features

  • 220B-parameter Australian model. Trellis Data says Agentaus 1 has about 220 billion parameters, roughly 50% more than its previous model, and is tuned for accuracy, multi-step reasoning and longer, more coherent outputs. Official source 1 ↗
  • Data kept in Australia. Prompts, files and outputs are stored and processed in Australian data centres on Trellis Data's own cloud rather than AWS or Azure, and are never used for training. Official source 1 ↗
  • Answers from your documents. Upload Word, PDF, Excel, PowerPoint and other common files, then search, summarise and compare them. Answers show their sources, with a reference for each response. Official source 1 ↗ · Official source 2 ↗
  • Web search and image analysis. Agentaus can search the web and read pages. Over the API it has four built-in tools: web search, web fetch, search over your own datasets, and image analysis with a vision model. Official source 1 ↗ · Official source 2 ↗
  • Protected and External modes. Protected Mode keeps everything in Australia. External Mode sends prompts to global engines such as Claude and ChatGPT using included credit, with the results kept protected. Official source 1 ↗
  • Team controls. Team and Business plans add shared sessions and documents, role-based (and on Business, attribute-based) access control, SSO with Active Directory, audit logs and spend limits. Official source 1 ↗ · Official source 2 ↗
  • Multilingual. You can ask, answer and translate in English, Indonesian, Chinese, Arabic and many more languages. Official source 1 ↗
  • OpenAI-compatible API. Call Agentaus from your own code through the OpenAI chat completions format, with tool calling, structured JSON output and a reasoning effort setting. Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗

Pricing & free limits

Free

A$0Per month

Protected Mode with a daily allowance of protected prompts, External Mode with a monthly top-up of Claude and ChatGPT credit, and Free-tier Agentaus agents. Official source 1 ↗

Standard

A$24Per month, excl. GST

5× the free tier's protected prompts, monthly prepaid Claude and ChatGPT credit with top-ups, and all Open-tier Agentaus agents. Official source 1 ↗

Professional

A$107Per month, excl. GST

Unlimited protected prompts, a large monthly Claude and ChatGPT credit, private agents you build and deploy, and corporate system integrations. Official source 1 ↗

Team

A$24Per seat / month; 2–19 seats

Everything in Standard plus central billing, shared sessions and documents (beta), role-based access control, SSO with Active Directory and audit logs. Official source 1 ↗

Business

A$107Per seat / month; 2–999 seats

Everything in Professional with unlimited protected prompts, role- and attribute-based access control, SSO, audit logs and spend limits per user and organisation. Official source 1 ↗

Enterprise / On Prem

CustomMinimum 1-year term

Deployed on premises, in your cloud or fully air-gapped, including at the edge, for Defence and high-security classifications. Billed at 200+ Business seats. Official source 1 ↗

Billing details

All prices are in Australian dollars and exclude GST. Official source 1 ↗

Paid personal plans include prepaid Claude and ChatGPT credit for External Mode, and you can top up extra credit whenever you need it. Official source 1 ↗

Enterprise and on-premises deployments need a minimum 1-year term and are billed at 200 or more Business seats. Official source 1 ↗

After a subscription ends there is a 7-day reactivation window. After that, the account's data is permanently deleted. Official source 1 ↗

Free access & limits

The free plan has a daily allowance of protected prompts and a monthly top-up of Claude and ChatGPT credit. Private agents and corporate integrations are not included. Official source 1 ↗

Best for & limitations

Turn a pile of policy submissions into a brief that sets out themes, concerns and disagreements, with a reference for every claim. Official source 1 ↗

Reuse past tender answers, or check contracts and quality manuals against updated rules or standards, with each gap flagged and referenced. Official source 1 ↗

Use AI on sensitive government, defence or legal work where data must stay in Australia, including offline on secure networks. Official source 1 ↗ · Official source 2 ↗

  • Benchmark claims are Trellis Data's own. Trellis Data says Agentaus 1 ranks 43rd globally among foundation models, but it doesn't name the benchmark or publish independent results. Official source 1 ↗
  • No open weights or model details. Agentaus 1 is only available through the Agentaus app and API. Trellis Data says its models are customised open-source models hosted on its own infrastructure, but it doesn't name the base model. Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗
  • Essential Eight level is self-assessed. Trellis Data holds ISO/IEC 27001, but its Essential Eight Maturity Level 2 claim is a self-assessment, with an independent assessment still to come. It also doesn't disclose data centre operators or exact locations. Official source 1 ↗
  • Limited client support for tool calling. Over the API, tool calling currently works with cURL and OpenCode; other clients such as Hermes Agent and Pi are planned. Official source 1 ↗
  • Answers can still be wrong. Trellis Data recommends checking important outputs before relying on them. Official source 1 ↗
Agentaus 1 product screenshot
Website previewFixed desktop viewport · 1440 × 900

Similar tools

Published tools that share its categories.

AskMedAI medical reference that answers clinical questions with citations to journals, guidelines and drug labels, for clinicians and students.
Research & DataPaidChecked Nov 7
Visit
SparkBotA Chrome extension that answers your saved questions about web pages and PDFs, quoting the line behind each answer.
Research & DataFree planChecked Oct 31
Visit