★ An Australian Frontier-AI Initiative

Australia can have sovereign frontier AI — without a billion-dollar moonshot.

Frontier-class capability no longer requires training a frontier model from scratch. It can be orchestrated — and we've built a working proof that lifts open models up a full tier. Here are the three paths to an Australian frontier model, and where we start.

Why this matters

Sovereign AI is national infrastructure.

The nations that control their own frontier-class AI control a strategic capability — for science, defence, industry and independence from foreign model providers and their export controls. The conventional assumption is that this means spending billions to pretrain a frontier model. It doesn't have to. A second route — orchestration — is now proven to reach frontier-class results, and it's a route an Australian lab can take today.

The landscape

Three paths to a frontier-class Australian model.

Ranked from hardest to most achievable. The capability at the end of each is comparable; the cost, time and risk are not.

1 · Train a sovereign frontier model from scratch moonshot

Build a GPT/Claude/Gemini-class foundation model domestically. This is the "true sovereignty" path — and it's brutally hard: a fleet of 100,000+ accelerators, vast proprietary data, one of the few dozen world-class pretraining teams on Earth, 12–18 month cycles, and a target that keeps moving. A handful of organisations globally can do it.

Cost: $ billionsTime: yearsTalent: scarcest in the worldRisk: very high

2 · A learned coordinator (trained orchestration) middle ground

Train a coordinator model that learns to decompose a task, route sub-tasks across a pool of models (including itself, recursively), and synthesise the result — all behind a single API. A handful of labs have already used this to match frontier benchmarks. Far cheaper than path 1 (no frontier pretraining), but you still train a model — and the result depends on the models in the pool.

Cost: moderateTime: monthsTalent: strong ML teamRisk: medium

3 · A hard-coded coordinator (engineered orchestration) fastest

A hand-engineered orchestrator: decompose → route each sub-task to the best available model → verify with multi-agent debate / best-of-N → aggregate. No model training — pure systems engineering with today's tooling. It reaches frontier-ish output by coordinating existing models and verifying hard. The quickest route to frontier-class capability.

Cost: lowTime: weeksTalent: strong engineeringRisk: low

★ Our proof-of-concept — one layer down we built this

Before scaling paths 2–3 to frontier, we prove the technique works one layer down: take fairly non-frontier, open-weight models and orchestrate them up to near-frontier performance. A lower starting point, chosen on purpose — so it's cheap to run and clean to verify.

Cost: ~$0 (open / free models)Time: nowVerifiable: yes — see below
The proof-of-concept

Lift open models up a full tier — and prove it.

We take open models and push them up to near-frontier capability — one deliberate layer down from the frontier itself. Why a layer down? Because at this level the result is both verifiable and immediately useful, not just a benchmark claim.

① It's objectively measured — no model judges another

We score on a public benchmark (MMLU-Pro) against the objective answer key — no model judging another. The honest finding: across a panel of free open models the correct answer is present 89.7% of the time — a selection ceiling +7.7 points above the best single model, and above a current frontier model. The winning knowledge is provably in the panel; capturing it reliably by coordination is the open frontier. Full method, every number, and the negative results: the research paper.

② We actually benefit

The proof is the product: a bootstrapped, ~$0 higher-class model built from free / open weights via coordination. A capability we can then use across everything we run — not a slide, a working asset.

③ It establishes the foothold

A working demonstration that orchestration lifts models a full tier is the credible foundation for the national pitch: scale the same method (paths 2–3) with real resources, and Australia has sovereign frontier-class AI capability.

④ It's honest

We don't claim to have pretrained a frontier model. We claim — and demonstrate — that coordination lifts a tier. That's a real, defensible, repeatable result, and the right first step on a serious path.

The result, in one chart

The winning answer is already in the crowd — a +7.7 ceiling.

73.0
Best single model,
alone
80.6
The same models,
voted together

Accuracy on MMLU-Pro, scored against the objective answer key. The selection ceiling — the score if you always picked the model that's right — sits +7.7 above the best single model, above the frontier target. Simple voting doesn't capture it yet; synthesis is the lever. See the full paper →

The manoeuvre, stated plainly:  a panel of free open models collectively holds the right answer far more often than any one of them — a +7.7 ceiling above the best single, above the frontier — and the open engineering problem is coordinating them to capture it. Measured honestly against an objective answer key, then scaled up with real resources.
Where this goes

From a proof to a sovereign capability.

The proof-of-concept de-risks the whole thesis. With the technique demonstrated, the path to an Australian frontier-class model is an engineering and resourcing question — not a moonshot. We're building the proof now; the national-scale version is paths 2 and 3 with the backing to match. If you're working on Australian sovereign AI capability and this is your problem too, that's exactly the conversation to have.