Xora

Overview

Xora is the reasoning model family that carried Infinity Intelligence through four generations of frontier work, from initial development in 2021 through to Xora 4 in 2026. It was designed for environments where a wrong answer is more expensive than a slow one: research, defense, and institutional deployments operating under strict data controls.

Like every Xora generation, it is a closed-weight model trained with grounding protocols that bias it toward saying less rather than guessing. That trade-off is deliberate, and it is the reason the family found the users it did.

Xora 4 is the family's final generation, and the one detailed below.

What shipped

Xora 4 Flash is the final released model in the family. It is the fast, lower-cost configuration of the generation, intended for high-volume institutional workloads where latency and predictable cost matter.

It remains supported for organizations already running it. No new capability work is planned.

What was cancelled

Xora 4 Reasoning was cancelled before release. It was the heavy configuration of the generation, and the one that carried most of the remaining research cost.

There will be no Xora 5, no Xora 4.1, no Xora 4 Code, and no further variants of any kind. The family is closed.

Support and availability

Xora 4 Flash is not being offered to new customers. Organizations with existing deployments keep them, and we continue to support those installations.

Teams looking for ongoing model development should look at PolyCode, or at the WillMe GPT models developed through Infinity Intelligence Labs.