BigBang-v1-MLX-oQ4e

This model was converted to MLX format from endless-frontier/BigBang-v1 using oMLX oQ quantization.

Quantization: oQ4e (enhanced 4-bit, data-driven mixed-precision via calibration). Uses imatrix-weighted quantization with GPTQ-based weight optimization for maximum quality retention at ~4.6 bpw.

Use with mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/BigBang-v1-MLX-oQ4e")

messages = [{"role": "user", "content": "Hello!"}]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
response = generate(model, tokenizer, prompt=prompt, max_tokens=512)

Or via CLI:

python3 -m mlx_lm.chat --model mlx-community/BigBang-v1-MLX-oQ4e

Use with oMLX

omlx import mlx-community/BigBang-v1-MLX-oQ4e

Model Info

• Architecture: qwen3_5_moe — 35B total, 3B active, 256 experts (8/tok)
• Attention: Hybrid linear + full attention
• Context: 262K tokens
• Type: Text-only (language model)

License

Apache 2.0 — Copyright 2026 Alibaba Cloud (original base model).