baa-ai/GLM-5.1-RAM-420GB-MLX

GLM-5.1 — 420GB (MLX)

Mixed-precision quantized version of zai-org/GLM-5.1 optimised by baa.ai using a proprietary Black Sheep AI method.

Per-tensor bit-width allocation via advanced sensitivity analysis and budget-constrained optimisation — no calibration data required.

Sized for 512GB Apple Silicon machines (378 GB model + 134 GB headroom for OS/KV cache).

Metrics

Metric	Value
Size	382 GB
Average bits	3.6
Architecture	MoE (256 experts)

Usage

from mlx_lm import load, generate

model, tokenizer = load("baa-ai/GLM-5.1-RAM-420GB-MLX")
response = generate(model, tokenizer, prompt="Hello!", max_tokens=256)
print(response)

---

Quantized by baa.ai