Which AI models can the AMD Radeon RX 9070 run?
Every model we track, checked against the AMD Radeon RX 9070 at 8K context. The largest that runs well is gpt-oss 20B.
VRAM16 GBAMD
Bandwidth640GB/s
Runs well28/69models at Q4_K_M or better
Fastest model~341tok/s · Qwen3.6 35B-A3B
Every model on the AMD Radeon RX 9070
Bars show memory used out of 16 GB. Pick a row for the full breakdown.
28 Runs well
NameMemory usedBest quantSpeed
gpt-oss 20B21BQ5_K_M · 15.7 GB~150 tok/s
Qwen3 14B15BQ6_K · 14.1 GB~32 tok/s- Qwen2.5 Coder 14B15BQ6_K · 14.3 GB~32 tok/s
DeepSeek R1 Distill Qwen 14B15BQ6_K · 14.3 GB~32 tok/s- Qwen2.5 14B15BQ6_K · 14.3 GB~32 tok/s
Phi-4 14B15BQ6_K · 14.3 GB~32 tok/s- Phi-4 Reasoning 14B15BQ6_K · 14.3 GB~32 tok/s
Mistral Nemo 12B12BQ8_0 · 14.9 GB~30 tok/s
Gemma 3 12B12BQ8_0 · 14.4 GB~30 tok/s- Gemma 4 12B12BQ8_0 · 13.7 GB~30 tok/s
- Gemma 2 9B9.2BQ8_0 · 12.5 GB~39 tok/s
- Qwen3 8B8.2BQ8_0 · 10.5 GB~44 tok/s
- DeepSeek R1 0528 Qwen3 8B8.2BQ8_0 · 10.5 GB~44 tok/s
Granite 3.3 8B8.2BQ8_0 · 10.6 GB~44 tok/s
Llama 3.1 8B8BQ8_0 · 10.2 GB~45 tok/s- Qwen2.5 7B7.6BQ8_0 · 9.2 GB~47 tok/s
- Qwen2.5 Coder 7B7.6BQ8_0 · 9.2 GB~47 tok/s
- DeepSeek R1 Distill Qwen 7B7.6BQ8_0 · 9.2 GB~47 tok/s
- Mistral 7B v0.37.3BQ8_0 · 9.4 GB~50 tok/s
- Gemma 3 4B4.3BFP16 · 9.5 GB~45 tok/s
- Qwen3 4B4BFP16 · 9.8 GB~48 tok/s
- Phi-4 Mini 3.8B3.8BFP16 · 9.4 GB~50 tok/s
- Phi-3.5 Mini 3.8B3.8BFP16 · 11.5 GB~50 tok/s
- Llama 3.2 3B3.2BFP16 · 8 GB~60 tok/s
- Qwen3 1.7B2BFP16 · 5.6 GB~95 tok/s
- Llama 3.2 1B1.2BFP16 · 3.3 GB~155 tok/s
- Gemma 3 1B1BFP16 · 2.6 GB~192 tok/s
- Qwen3 0.6B0.8BFP16 · 3 GB~256 tok/s
20 Fits with heavy quantization
NameMemory usedBest quantSpeed
- Qwen3.6 35B-A3B36BQ2_K · 14.3 GB~341 tok/s
- Qwen2.5 32B33BQ2_K · 15 GB~31 tok/s
- DeepSeek R1 Distill Qwen 32B33BQ2_K · 15 GB~31 tok/s
- Qwen3 32B33BQ2_K · 15 GB~31 tok/s
- QwQ 32B33BQ2_K · 15 GB~31 tok/s
- Qwen2.5 Coder 32B33BQ2_K · 15 GB~31 tok/s
Nemotron 3 Nano 30B-A3B32BQ2_K · 12.5 GB~293 tok/s- Gemma 4 31B31BQ2_K · 13.5 GB~33 tok/s
GLM-4.7 Flash 30B-A3B31BQ2_K · 12.7 GB~341 tok/s- Qwen3 30B-A3B31BQ2_K · 12.8 GB~310 tok/s
- Qwen3 Coder 30B-A3B31BQ2_K · 12.8 GB~310 tok/s
- Qwen3.8 27B28BQ3_K_M · 14.7 GB~28 tok/s
- Qwen3.6 27B28BQ3_K_M · 14.7 GB~28 tok/s
- Gemma 3 27B27BQ3_K_M · 15.1 GB~29 tok/s
- Gemma 2 27B27BQ2_K · 13.1 GB~38 tok/s
- Gemma 4 26B-A4B26BQ3_K_M · 13.5 GB~207 tok/s
- Mistral Small 3.2 24B24BQ3_K_M · 13.7 GB~33 tok/s
- Mistral Small 24B24BQ3_K_M · 13.5 GB~33 tok/s
- Devstral Small 24B24BQ3_K_M · 13.5 GB~33 tok/s
- Magistral Small 24B24BQ3_K_M · 13.5 GB~33 tok/s
21 Too big
NameMemory usedBest quantSpeed
Kimi K3.1 Pro 2.8T-A104B2800Bneeds 1698.3 GB16 GB free- Qwen3.8 2.4T-A95B2446Bneeds 1484.3 GB16 GB free
- DeepSeek V4 Pro 1.6T-A49B1600Bneeds 971.2 GB16 GB free
- Mistral Large 4 1T-A52B1000Bneeds 607.5 GB16 GB free
- DeepSeek V4.1 Flash763Bneeds 463.4 GB16 GB free
- GLM-5.3753Bneeds 457.8 GB16 GB free
- GLM-5 744B-A40B744Bneeds 452.4 GB16 GB free
MiniMax M3 428B-A23B427Bneeds 260.5 GB16 GB free- GLM-5.3 Flash 320B-A18B321Bneeds 195.3 GB16 GB free
- DeepSeek V4 Flash 284B-A13B284Bneeds 173 GB16 GB free
- Qwen3 235B-A22B235Bneeds 144.7 GB16 GB free
- Qwen3.8 Flash Next180Bneeds 109.9 GB16 GB free
- gpt-oss 120B117Bneeds 71.7 GB16 GB free
- Llama 4 Scout 17B-16E109Bneeds 68.3 GB16 GB free
- GLM-4.5-Air 106B-A12B106Bneeds 66.4 GB16 GB free
- Qwen3 Coder Next 80B-A3B80Bneeds 49.1 GB16 GB free
- Qwen2.5 72B73Bneeds 47.4 GB16 GB free
- Llama 3.1 70B71Bneeds 46.1 GB16 GB free
- Llama 3.3 70B71Bneeds 46.1 GB16 GB free
- DeepSeek R1 Distill Llama 70B71Bneeds 46.1 GB16 GB free
- Mixtral 8x7B47Bneeds 30 GB16 GB free
Run it in the cloud
Need more than 16 GB? Rent a bigger GPU by the hour.
Rent a cloud GPU by the hour. No upfront cost, stop any time.