Ornith-1.5-35B-A3B

Overview

The official BF16 release of Ornith's mid-size MoE - activates ~3B of 35B parameters per token while significantly outperforming Qwen3.6-35B across coding and agentic benchmarks. 262K context; for vLLM and SGLang.

Good for

• agentic
• chat
• coding

Recommended hardware

≈ 96 GB VRAM to run this 35B model at the BF16 quantization.

Needs 128 GB+ of VRAM or unified memory (e.g. multi-GPU with 2x 96 GB cards, or a 128 GB Mac)

Links

• View on HuggingFace
• Run on Infersec