XNews

AI model/product highlights: SenseNova U1 Pro and Grok parameter discussion

AI model/product highlights: SenseNova U1 Pro and Grok parameter discussion

SenseTime unveiled SenseNova U1 Pro flagship with native 8K output and dense text handling; separate posts compare Grok and other model parameter counts and cost

Key insights

SenseTime released SenseNova U1 Pro: native 8K and dense text fidelity

SenseTime introduced the SenseNova U1 Pro as the flagship of the 'SenseNova U' series, showcased at WAIC on July 18. The U1 Pro emphasizes four capabilities: professional design aesthetics, native 8K resolution output, high-fidelity handling of dense layouts and small text without blurring, and the ability to run dozens of generation iterations around a single target. The announcement positions the U1 Pro as focused on complex page layouts and repeated generation cycles to refine outputs.

Parameter and cost comparisons: Grok 4.5, Kimi K3, and efficiency claims

A separate post compared model sizes and costs: Grok 4.5 was stated to have 1.5 trillion parameters. Kimi K3 was stated to have 2.8 trillion parameters but to incur roughly three times higher per-task cost. The post also asserted that SpaceXAI achieved improvements in 'intelligent efficiency' and suggested imagining a 30-trillion-parameter Grok as a rhetorical device to consider potential capability increases.

Source posts