AMD · RDNA 2 · 8 GB VRAM tier
Best AI Models for AMD Radeon RX 6600M (8 GB)
Runs 7–9B models well; 14B doesn't fit in 8 GB.
8 GB GDDR6224 GB/s18 TFLOPS FP16100 WRDNA 2 2021
8 GB at 224 GB/s: ~7-8B Q4 models, no more. Not on AMD's official ROCm list; the Vulkan backend (llama.cpp, LM Studio) is the easy path.
| # | Model | Speed | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 9BAlibabaHow to run → Q4_K_M · 6.6 GB · 21.1 t/s | 21.1 tok/s | Q4_K_M · 6.6 GB | |
| 2 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 41.2 t/s | 41.2 tok/s | Q4_K_M · 3.4 GB | |
| 3 | Gemma 2 9B ITGoogle Q4_K_M · 7.3 GB · 22.0 t/s | 22.0 tok/s | Q4_K_M · 7.3 GB | |
| 4 | Gemma 3 4B ITGoogle Q4_K_M · 5.0 GB · 47.3 t/s | 47.3 tok/s | Q4_K_M · 5.0 GB | |
| 5 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB · 85.1 t/s | 85.1 tok/s | Q4_K_M · 5.5 GB |
| # | Model | Speed | Reasoning Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 9BAlibabaHow to run → Q4_K_M · 6.6 GB · 21.1 t/s | 21.1 tok/s | Q4_K_M · 6.6 GB | |
| 2 | Gemma 3 4B ITGoogle Q4_K_M · 5.0 GB · 47.3 t/s | 47.3 tok/s | Q4_K_M · 5.0 GB | |
| 3 | Qwen2.5 7B InstructAlibaba Q4_K_M · 5.8 GB · 26.9 t/s | 26.9 tok/s | Q4_K_M · 5.8 GB | |
| 4 | Gemma 2 9B ITGoogle Q4_K_M · 7.3 GB · 22.0 t/s | 22.0 tok/s | overall | Q4_K_M · 7.3 GB |
| 5 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB · 85.1 t/s | 85.1 tok/s | Q4_K_M · 5.5 GB |
| # | Model | Speed | Long context Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 4BAlibabaHow to run → Q4_K_M · 3.4 GB · 41.2 t/s | 41.2 tok/s | Q4_K_M · 3.4 GB | |
| 2 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB · 21.1 t/s | 21.1 tok/s | Q4_K_M · 6.6 GB | |
| 3 | Gemma 4 E4B ITGoogle Q4_K_M · 5.6 GB · 25.3 t/s | 25.3 tok/s | Q4_K_M · 5.6 GB | |
| 4 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB · 85.1 t/s | 85.1 tok/s | Q4_K_M · 5.5 GB | |
| 5 | MiniCPM5 2BOpenBMB Q4_K_M · 2.5 GB · 70.7 t/s | 70.7 tok/s | Q4_K_M · 2.5 GB |
| # | Model | Speed ↓ | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 4BAlibabaHow to run → Q4_K_M · 3.4 GB · 41.2 t/s | 41.2 tok/s | Q4_K_M · 3.4 GB | |
| 2 | Ling 3.0 TinyInclusion AI Q4_K_M · 5.5 GB · 85.1 t/s | 85.1 tok/s | Q4_K_M · 5.5 GB | |
| 3 | MiniCPM5 2BOpenBMB Q4_K_M · 2.5 GB · 70.7 t/s | 70.7 tok/s | Q4_K_M · 2.5 GB | |
| 4 | Gemma 4 E4B ITGoogle Q4_K_M · 5.6 GB · 25.3 t/s | 25.3 tok/s | Q4_K_M · 5.6 GB | |
| 5 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB · 21.1 t/s | 21.1 tok/s | Q4_K_M · 6.6 GB |
| # | Model | Speed | Vision Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.5 9BAlibabaHow to run → Q4_K_M · 6.6 GB · 21.1 t/s | 21.1 tok/s | Q4_K_M · 6.6 GB | |
| 2 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 41.2 t/s | 41.2 tok/s | Q4_K_M · 3.4 GB | |
| 3 | Gemma 3 4B ITGoogle Q4_K_M · 5.0 GB · 47.3 t/s | 47.3 tok/s | Q4_K_M · 5.0 GB | |
| 4 | Q4_K_M · 6.5 GB · 21.6 t/s | 21.6 tok/s | Q4_K_M · 6.5 GB | |
| 5 | MiniCPM V 4.7 1BOpenBMB Q4_K_M · 1.6 GB · 105 t/s | 105 tok/s | Q4_K_M · 1.6 GB |
See all 218 models on the RX 6600M →Best quant that fits · speeds estimated on this card
The RX 6600M runs 7–9B models well. RDNA 2 generation, released 2021.
GenerationRDNA 2released 2021
RX 6600M strengths and weaknesses
Strengths
- Runs 7–9B models: Qwen3.5 9B at 21.1 tok/s in 6.6 GB
- Low power draw: 100 W board power; a 200 W PSU is enough
Weaknesses
- 14B models don't fit: Qwen3 14B needs 11.9 GB at Q4 vs 8 GB here
- Little room for context: Qwen3.5 9B uses 6.6 of 8 GB, 1.4 GB left
- Slow prompt reading: A 4K-token prompt on Qwen3.5 9B waits 20.5 s for the first token
- ROCm, not CUDA: No FP8 weights; MXFP4 only through llama.cpp, not vLLM
Cards people weigh against the RX 6600M, each with the one reason it is worth a look. Speeds on Qwen3.5 9B.
39% faster, same 8 GBRTX 4060 Ti 8GBNVIDIA · 8 GB · Ada Lovelace
▲ 39% speedsame 8 GBabout $399rent $0.19/hr
4 GB more and 74% fasterRTX 3060 12GBNVIDIA · 12 GB · Ampere▲ 74% speed▲ 4 GBabout $329rent $0.06/hr
4 GB more and nearly 2× the speedRX 9070 GREAMD · 12 GB · RDNA 4▲ 93% speed▲ 4 GBabout $549
32% faster, same 8 GBRTX 4060NVIDIA · 8 GB · Ada Lovelace▲ 32% speedsame 8 GBabout $520rent $0.08/hr
8 GB more and 3.5× the speedRTX 4080NVIDIA · 16 GB · Ada Lovelace▲ 247% speed▲ 8 GBabout $1,199rent $0.25/hr
8 GB more and 3.6× the speedRTX 4080 SUPERNVIDIA · 16 GB · Ada Lovelace▲ 256% speed▲ 8 GBabout $999rent $0.32/hr
Output speed depends on memory bandwidth (224 GB/s). Grey rows need more than 8 GB.
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB21.1tok/s
- 14BQwen3 14B · not at 4-bitneeds 11.9 GB
- 27–31BQwen3.8 27B · not at 4-bitneeds 17.9 GB
- 35B MoEQwen3.6 35B A3B · not at 4-bitneeds 22.0 GB
- 70BLlama 3.3 70B Instruct · not at 4-bitneeds 51.3 GB
- 120B MoEGPT OSS 120B · not at 4-bitneeds 70.8 GB
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
All 218 models on the RX 6600M
Featured onlyBest quant that fits, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
CodeQwen1.5 7BChatCode | Q4_K_M | 5.7 GB | 28.1 t/s | 66K | A69, Runs well | |
Xgen 7B 8k BaseChat | Q4_K_M | 5.7 GB | 29.1 t/s | 8K | A70, Runs well | |
Yi 9BChat | Q4_K_M | 6 GB | 23.2 t/s | 4K | B64, Decent | |
| Q4_K_M | 5.0 GB | 29.1 t/s | 4K | A70, Runs well | ||
Qwen 1 8BChat | Q4_K_M | 1.9 GB | 111.1 t/s | 8K | S93, Runs great | |
Baichuan 7BChat | Q4_K_M | 5.0 GB | 29.1 t/s | 4K | A70, Runs well | |
Yi 6B ChatChat | Q4_K_M | 4.2 GB | 33.0 t/s | 4K | A72, Runs well | |
| Q4_K_M | 4.6 GB | 29.1 t/s | 2K | A70, Runs well | ||
Phi 4 ReasoningChatMathCodeReasoning | IQ2_XXS | 7.7 GB | 28.3 t/s | 33K | C44, Usable | |
Qwen2.5 Omni 7BChat | Q3_K_S | 8.0 GB | 26.0 t/s | — | C36, Usable | |
Qwen2.5 Coder 3BChatCodeFunctions | Q4_K_M | 2.8 GB | 60.3 t/s | 33K | A84, Runs well | |
Qwen 7BChat | Q4_K_M | 7.6 GB | 26.4 t/s | 33K | C42, Usable | |
MiniCPM V 2 6Vision | Q4_K_M | 7.9 GB | 25.1 t/s | — | C36, Usable | |
Llama 3.2 11B Vision InstructVisionFunctions | Q2_K | 7.8 GB | 26.9 t/s | — | C41, Usable | |
Gemma 3 12B ITVision | IQ2_M | 7.4 GB | 29.7 t/s | 33K | B52, Decent |
Specs
- Architecture
- RDNA 2
- VRAM
- 8 GB GDDR6
- Memory bandwidth
- 224.0 GB/s
- FP16 (dense)
- 17.55 TFLOPS
- Stream processors
- 1,792
- TDP
- 100 W
- Released
- 2021-05-31
RX 6600M FAQ
What is the largest model the RX 6600M can run?
Qwen3.5 9B (9.7B) is the largest at 4-bit or better: at Q4_K_M it uses 6.6 GB of 8 GB, and runs at about 21.1 tok/s.
Can the RX 6600M run 14B models?
No. Qwen3 14B needs 11.9 GB at Q4, more than the 8 GB on this card.
How many tokens per second does the RX 6600M give?
About 21.1 tok/s on Qwen3.5 9B at Q4_K_M, with the first token after 20.5 s on a 4K-token prompt.
What power supply does the RX 6600M need?
Plan a 200 W PSU (2× the 100 W TDP).
Is renting cheaper than buying an RX 6600M?
An RTX 3090 (24 GB) rents from $0.18/hr. With regular daily use, owning pays off.