StudioTether releases BotPsy, a 0.8B on-device agent model built on Qwen3.5-0.8B
Original title: Stablecoin issuer Tether releases BotPsy, an AI agent small enough for a phone
Tether AI Research released BotPsy, a 0.8B-parameter agent language model built on Qwen3.5-0.8B, designed to make tool calls for on-device tasks like messaging and calendar management. Its weights are downloadable on Hugging Face under Apache 2.0 (with a "research and educational use" note), GGUF files range from 529 MB (Q4_K_M) to 1.52 GB (BF16), and Tether's own tests show it averaging 45.0 across six tool-use benchmarks versus 24.7 for its base model and 38.8 for Qwen3.5-2B, while trailing on general knowledge tests like MMLU-Pro and GPQA Diamond; no phone speed figures are provided and prompt-injection unsafe answers still occur 31% of the time.
The piece aggregates official benchmark scores, quantized file sizes and license caveats, and notes the results come from Tether's own tests, helping readers weigh this on-device agent model.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.