# Radeon RX 9070 XT for local AI

URL: https://insidethemachine.ai/machines/radeon-rx-9070-xt
Type: AMD graphics card. Chip: Navi 48. Memory: vram, options 16 GB. Bandwidth: 640 GB/s. Spec source: https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units (2026-10-15).
Price: from about $760 (2026-10-15, approximate). Status: current.

## Which models fit

| memory GB | usable GB | models that fit | best 3 |
|---|---|---|---|
| 16 | 15 | 44 | K2 Horizon 7B (8.6 GB), Gemma 4 12B (11.2 GB), Qwen3.5 9B (9.6 GB) |

## Published speeds (not our tests)

- gpt-oss-20b: 140 tok/s decode (llama.cpp, unspecified, , community-unverified) https://llm-speed.com/m/gpt-oss-20b (2026-10-15)

## Questions

**What is the best local LLM for the Radeon RX 9070 XT?**
With 16 GB, K2 Horizon 7B (needs 8.6 GB, AA 20.6). 44 of 124 models fit.

**How fast is the Radeon RX 9070 XT for local AI?**
Published: gpt-oss-20b at 140 tokens/sec (llama.cpp, ). These are other people's measurements, not ours.

**How much does the Radeon RX 9070 XT cost?**
From about $760 (as of 2026-10-15). 2026 prices are inflated by the memory-chip shortage; check before buying.