A Guth Labs publication ↗
Back to News

Phancy announces Phai-A1 language model and Phai-M1 video model

Green glass orb encircled by a cream ribbon and linked to a blank cream card with gold light
AI-generated editorial illustration

Phancy announced Phai-A1, a language model for coding and AI assistants, and Phai-M1, a video-generation model, on November 18 in a company release distributed by Media OutReach.

Phai-A1 supports a native one-million-token context window and tasks including code generation, bug fixing, tool use and multistep workflows. The company says it can run locally on a personal computer, depending on the available hardware. It did not provide minimum hardware requirements in the announcement.

Phai-M1 targets live-action-style film and television content, generating video and synchronized audio together. Phancy says the model is designed to improve expressions, movement and visual detail. Those are company claims, not independent quality measurements.

The models are integrated with Phancy products including PhanthyCode and PhanthyMovie. The company also says it optimized the Phai series for Huawei’s Ascend computing platform.

Phancy plans to release both models’ weights in the near future; the announcement does not establish that downloadable weights are already available. It directs users to its Phai service to try the models.

The planned weight release is the key next step for developers assessing deployment flexibility, rather than relying solely on the company’s hosted applications.

Sources and citations

Revision history

Published .