# Ornith 1.5 35B MoE

> Canonical: https://www.overmindlab.ai/models/ornith-1-5-35b-a3b

35B MoE Qwen 3.5 finetune, 3B active, with a 256K window.

| Property | Value |
| --- | --- |
| Model ID | ornith-ai/Ornith-1.5-35B-A3B |
| Provider | Ornith |
| Group | Ornith |
| Parameters | 35B MoE (3B active) |
| Tier | Large |
| Context window | 256K |
| Max training context | 256K |
| Tool calling | Yes |
| Training methods | LoRA |
| Training cost | from $1.52 |
| Serving cost (per 1M output tokens) | from $8.00 |

## Overview

Same architecture as Qwen 3.5 35B-A3B: 10 full-attention layers, 256 experts, 256K positions. LoRA uses that full window.

Ornith 1.5 35B-A3B is a Qwen 3.5 35B-A3B finetune. A 2-step LoRA completed on the Qwen 3.5 Unsloth image, and the pinned vLLM image serves it with thinking disabled.

## Good for

- Agents that start from the Ornith 1.5 MoE checkpoint
- LoRA on a 35B Qwen 3.5 mixture of experts
- Tool-calling rows

Train this model on your agent's data: https://docs.overmindlab.ai/models/training.md

## Related models

- [Ornith 1.5 9B](https://www.overmindlab.ai/models/ornith-1-5-9b): 9B, 256K context