Open-weight Qwen3.8 2.4T A95B. Same Intelligence 58 as Max at $2/$6. The weights NVIDIA documented for GB300 NVL72 serving.
“The open-weight sibling of Qwen3.8 Max.”
Self-host only with datacenter GPUs; otherwise use the Max API.
Training data policy: N/A — open weights. Self-host for full control.
DPDP Act (India): Self-host in India for full DPDP compliance.
Data residency: Self-hosted
Self-host in India