RX 7600 8GBBudget pick
C106.9 tok/s decode
~$269 MSRP
Can it run?
Runs with offload (needs ~0 GB host RAM)
Using Q6_K in Ollama
Fit status
Runs with offload (needs ~0 GB host RAM)
Decode
34.8 tok/s
TTFT
5557 ms
Safe context
16K
Memory
4.0 GB / 4.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs with offload (needs ~0 GB host RAM) | 34.8 tok/s | 8082 ms | 32K |
| Chat | C | Runs with offload (needs ~0 GB host RAM) | 34.8 tok/s | 3031 ms | 8K |
| Coding | C | Runs with offload (needs ~0 GB host RAM) | 34.8 tok/s | 5557 ms | 16K |
| RAG | C | Runs with offload (needs ~0 GB host RAM) | 34.8 tok/s | 10103 ms | 32K |
| Reasoning | C | Runs with offload (needs ~0 GB host RAM) | 34.8 tok/s | 6567 ms | 16K |
How gemma 2 2b it (2B params) fits at each quantization level on Intel Arc A370M 4GB (4.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.8 GB | Low | D34 |
Q3_K_S | 3 | 1.0 GB | Low | D35 |
NVFP4 | 4 | 1.1 GB | Medium | D35 |
Q4_K_M | 4 | 1.2 GB | Medium | D36 |
Q5_K_M | 5 | 1.4 GB | High | D37 |
Q6_KBest for your GPU | 6 | 1.6 GB | High | D38 |
Q8_0 | 8 | 2.1 GB | Very High | C41 |
F16 | 16 | 4.1 GB | Maximum | F0 |
huggingface-cli download hf-bartowski--gemma-2-2b-it-ggufUpgrade options
~$269 MSRP
Intel Arc A750 8GBBest value~$289 MSRP
GTX 1660 Super 6GBBiggest leap
Intel Arc Pro A40 6GBIntel upgrade