Ornith AI Releases DFlash Speculative Decoding Models for Ornith-1.5 Series

·2026.09.29 15:57

Key point

The release includes DFlash draft models for the 9B, 35B-A3B, and 397B variants of Ornith-1.5.

Details

Ornith AI has released DFlash draft models designed for speculative decoding with the Ornith-1.5 series. These models are available on Hugging Face for three specific parameter sizes:

Ornith-1.5-9B-DFlash: Pairs with the Ornith-1.5-9B model.
Ornith-1.5-35B-A3B-DFlash: Pairs with the Ornith-1.5-35B-A3B model.
Ornith-1.5-397B-DFlash: Pairs with the Ornith-1.5-397B model.

Speculative decoding typically accelerates inference by using a smaller draft model to propose tokens that the larger target model verifies in parallel.

llm model-release speculative-decoding inference-optimization hugging-face ornith-ai

Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.