Trellis Data announces 220-billion-parameter Agentaus 1
Trellis Data announced Agentaus 1 on 10 November 2026, describing it as its next-generation sovereign AI model. The company says the model ranks 43rd globally among foundational AI models.

Key takeaways · 5
- 01
Trellis Data announced Agentaus 1 on 10 November 2026 as its next-generation sovereign AI model.
- 02
The model has 220 billion parameters; Trellis Data says it is about 50 per cent larger than its previous model.
- 03
Access is offered through free, subscription, enterprise and API tiers.
- 04
Agentaus lists personal plans at $24 and $107 per month, excluding GST.
- 05
Trellis Data says Agentaus 1 can operate at scale on smaller compute clusters.
A larger sovereign model
Trellis Data announced Agentaus 1 on 10 November 2026, calling it a next-generation sovereign AI model.[1] It has 220 billion parameters, and the company says it is about 50 per cent larger than its previous model.[1] Trellis Data says Agentaus 1 ranks 43rd globally among foundational AI models.[1] The company says it designed the model for greater accuracy and deeper multi-step reasoning, as well as longer, more coherent outputs.[1]
Compute and task scope
Trellis Data says Agentaus 1 can operate at scale on smaller compute clusters.[1] It also says the model is designed to reduce the need to use multiple AI models for different tasks.[1] Trellis Data says the model demonstrates that world-class AI models can be developed in Australia.[1] Readers assessing it for work can distinguish these design claims from evidence about performance in their own tasks.
Access, pricing and data
Agentaus 1 is available to individuals and businesses through free, subscription, enterprise and API tiers.[1] Agentaus lists its Free plan at $0 per month, excluding GST, and its Standard and Professional personal plans at $24 and $107 per month, excluding GST.[3] The Team plan is listed at $24 per seat monthly for teams of two to 19 seats; Business is $107 per seat monthly for organisations with two to 999 seats, both excluding GST.[3] Enterprise, Bespoke and On Prem deployments have custom pricing, a minimum one-year term and billing at 200 or more Business seats.[3] Agentaus says its assistant works in secure Australian data centres under Australian law and does not use customer data to train AI.[3]
Australian focus and competition
Trellis Data says the model gives Australian businesses and government access to locally developed AI with security and sovereign data protection.[1] The company says it has spent a decade building applied AI capability in Australia, including for high-security environments, and that customers can get implementation and adoption support from its Canberra-based team.[1] Trellis Data says it intends to build trusted AI for Australian users before taking its AI expertise into international markets.[1] The Australian Financial Review reported that Trellis Data says Agentaus 1 could win business from major US AI companies; the company was backed by less than $10 million in external funding.[2] The AFR also reported that Anthropic was trying to convince Australian investors its AI platforms made it a $2 trillion company.[2]
For organisations evaluating AI, Agentaus 1 combines several access tiers with company-stated Australian data protections, so pricing, deployment needs and data handling are relevant points to assess. The announcement also puts Trellis Data’s claims about capability and local development into a market context that includes major US AI companies.
Why it matters
AI training for your team
kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.
How this developed
19 November 2026
Trellis Data announces 220-billion-parameter Agentaus 1
Sources
- Trellis Data joins global AI ranks with launch of near-frontier sovereign AI model – Agentaus 1 — Trellis Datatrellisdata.com
- New Australian AI model by Trellis Data, Agentaus 1, wants to challenge US tech giants like Anthropicafr.com
- Agentaus: Australia's Sovereign AI Platform | Trellis Datatrellisdata.com