You can deploy this model and start earning money today!

nathansutton/
Ornith-1.0-35B-Q6-MLX

Name: Ornith-1.0-35B-Q6-MLX
Availability: InStock

deepreinforce-ai/Ornith-1.0-35B

quantized

6-bit

Views

No views yet

Ornith-1.0-35B-Q6-MLX

An MLX quantization of deepreinforce-ai/Ornith-1.0-35B for running locally on Apple Silicon — a higher-bit build for big-memory Macs, used by the Terminal-Bench kit of chad, a Claude-Code-style local coding agent.

Quant

Params: 35B (MoE, 256 experts / top-8, ~3B active per token)
Scheme: uniform 6-bit group-64 affine (routers 8-bit, A_log fp32) — near-reference quality, same class as a llama.cpp Q6_K
Footprint: ~28.5 GB weights; needs a ≥48 GB Apple Silicon Mac (~36 GB Metal working set)
RAM: Does NOT fit 24 GB Macs — use the UD-Q2_K_XL repo there.

Use it

With chad (auto-downloads this model on first run):

uvx --from git+https://github.com/nathansutton/mlxcc chad

Or directly with mlx-lm:

python

from mlx_lm import load, generate
model, tok = load("nathansutton/Ornith-1.0-35B-Q6-MLX")
print(generate(model, tok, "Write a haiku about quantization.", max_tokens=64))

License

Inherits the license of the base model deepreinforce-ai/Ornith-1.0-35B. Review it before use or redistribution.