2026 for Qubic: peer review you cannot buy

wanna show you something the market hasn't seen yet.

In 2026, three Qubic researchers walked into IEEE conferences and got three papers accepted. In Berlin, a room full of academics - none of them holding a single Qubic token - voted their talk the best of the night.

You can buy influencers. You can buy stages. You cannot buy peer review. That's the whole point of this post.

Three names

David Vivancos has worked on machine intelligence since 1995. 26,000+ hours. Four hundred conferences.

José Sánchez García: doctorate, research position at UNIR, an accredited Spanish university. Published and cited.

The papers carry those two names. The third stands behind the thing the papers study: Come-from-Beyond, the engineer who conceived Aigarth more than twenty years ago and built toward it through NXT, IOTA, and Qubic. The architecture under review is his.

What actually happened this year

March. Two acceptance letters arrived within weeks of each other: the trinary paper for Osaka, the Neuraxon 2.0 paper for Berlin. Both IEEE. Before a single talk was given, the papers had already passed review. April added two more submissions - AGI-26 and ALIFE. Early May, a fifth went to NeurIPS.

Berlin, May. The first live presentation. Neuraxon 2.0 at the 11th ICMLT, an IEEE machine learning conference. Eight talks ran in the evening session. The researchers in the room voted Qubic's the best of the night. None of them had any connection to the project. That's what makes it worth anything.

Osaka, four days from now. July 22, 15:00, Kansai University. The trinary paper: a neuron with three states instead of two. On. Off. And deliberately silent. The claim: silence is a function, and it buys stability that binary systems pay for in waste.

San Francisco, July 27–30. The MultiNeuraxon paper is accepted at AGI-26, the only major conference series on earth dedicated entirely to artificial general intelligence. It enters the Springer proceedings, next to serious research, reviewed by people who owe Qubic nothing.

ARC-AGI-3, July. The benchmark built specifically to break LLMs: interactive puzzles, no instructions, nothing to memorize. Humans solve them easily. Every major AI model at launch scored under 1%. In six days and seven submissions, Neuraxon hit 0.18 - above Grok on that run, in the same near-zero band as the biggest labs on the planet.

It's a starting line. But it's the same starting line as Google's and OpenAI's, on the one test that measures learning instead of memorizing. A crypto project's evolved architecture is standing on it. That same week, the Anna training task converged to zero error.

NeurIPS, September 24. The fifth paper sits at the largest AI conference in the world - Sydney, December. It takes Spearman's g, the general factor of human intelligence measured since 1904, and asks whether it can be grown inside an evolved artificial brain. Reviews come back July 22. The decision lands September 24. Most NeurIPS submissions get rejected - the team published the date anyway. Remember it.

Who's in the room in San Francisco

Look at the speaker list. Karl Friston - the most cited neuroscientist of his generation, whose theory sits in the footnotes of Qubic's own research line. The poster will stand in a building where the author of its theoretical foundation gives a keynote. Michael Levin. Neil Gershenfeld of MIT. Gary Marcus. Ben Goertzel, the man who named the field. Past speakers of this series: two Turing Award winners, and François Chollet - who built the very ARC benchmark Neuraxon just walked onto.

And on July 30, AGI-26 runs a dedicated Investor Day. The AGI investment ecosystem in one room, days after Qubic's work enters the Springer record. I hope Qubic gets the attention there that the work has earned.

One more thing.

Everything above - the papers, the awards, the benchmarks - is scaffolding for a single word: Aigarth. AGI grown on a decentralized network. The idea Come-from-Beyond has carried for more than twenty years.

Anna reached zero error this month. The convergence method works. New benchmarks are already loading: the ARC-AGI leaderboard climb before San Francisco, Neuraxon 3.0 in Q3, the g-factor verdict in September.

We wish the Qubic team every success and thank them for their work in the Qubic family. Forward!

WE ARE STILL VERY EARLY.