Palmyra-mini: WRITER ships a 1.5B to 1.7B parameter model family
- Source
- x.com
- Date
Meet Palmyra-mini: our family of 1.5 to 1.7B parameter models that punch well above their weight class. Palmyra-mini acts as an everywhere-capable layer that handles the speed- and cost-sensitive parts of a workflow – while leaving deeper reasoning to larger models. Small enough to run almost anywhere. Smart enough to matter. Learn more about the Palmyra-mini family:

Context
WRITER positions Palmyra-mini as a small-model layer for work where speed and cost matter, leaving deeper reasoning to larger models. The family spans roughly 1.5 to 1.7 billion parameters. Its release article identifies content checks, scoring, routing and preliminary drafting as intended uses. This suggests a division of labor: use a compact model for frequent, bounded steps and reserve more demanding reasoning for another model. The company's reported throughput measurements used a server GPU, so they should not be read as demonstrated performance on every personal device.
Checking sign-in…
Loading comments…






