TestingCatalog describes it as a natively multimodal mixture-of-experts model with 1 trillion total parameters and 49 billion active parameters. The article says model weights are planned by the end of October; they are not part of the preview launch described here.
The article lists API pricing at $1.36 per million input tokens and $4.18 per million output tokens. It reports scores of 61.7% on DeepSWE v1.1 and 59.9% on AutomationBench, alongside results on other evaluations. These performance figures are claims reported by the source and should not be treated as independently verified comparative results. The concrete change is preview access now, with weights still pending.