Dev ports Inkling-2-Mini to a single AMD R9700: 196K context at ~110 tok/s

abhinand05 · reddit · 2026-11-22

A developer ported Inkling-2-Mini's official NVFP4/MXFP8 checkpoint to vLLM on an AMD R9700 (32GB, RDNA4) — upstream only supports NVIDIA and MI355X — releasing it as the open-source vllm-redshift project.

Key work and results:

The author plans day-one ports for upcoming releases like Qwen4, plus support for 24GB cards (7900 XTX) with custom quants.

Original post →

More from coding & agent

coding & agent channel →