NVIDIA GeForce GTX 1650
NVIDIA GeForce GTX 1650 is a entry-level GPU from NVIDIA. It offers 4 GB of VRAM, suitable for small and heavily quantized models. It delivers 2.98 TFLOPS of FP32 compute performance with 128 GB/s of memory bandwidth It can run 8 models in our database. It's a cost-effective entry point for experimenting with local LLMs.
Specifications
| Vendor | NVIDIA |
| Full Name | NVIDIA GeForce GTX 1650 |
| VRAM | 4GB |
| Performance Tier | Entry |
| Benchmark Score | 7,800 |
| FP32 Compute | 2.98TFLOPS |
| Memory Bandwidth | 128GB/s |
| Compatible Models | 8Models that can run on this GPU |
Strengths
- Affordable entry point for trying local LLMs
Limitations
- 4 GB VRAM — limited to small and heavily quantized models
- 2.98 TFLOPS — modest compute may limit inference speed
- 128 GB/s bandwidth — may bottleneck larger model inference
- Limited future-proofing — may struggle with next-generation models
Compatible Models (8)
Qwen68%
Qwen3 4B Q4_K_M
4.0B
3.4 GB
Q4_K_M
32,768 ctx
Gemma66%
Gemma 3 4B Q4_K_M
4.0B
3.4 GB
Q4_K_M
32,768 ctx
DeepSeek56%
DeepSeek R1 Distill Qwen 1.5B Q4_K_M
1.5B
1.3 GB
Q4_K_M
131,072 ctx
Qwen55%
Qwen3 1.8B Q4_K_M
1.8B
1.5 GB
Q4_K_M
32,768 ctx
Gemma52%
Gemma 3 1B FP16
1.0B
2.6 GB
FP16
32,768 ctx
Gemma50%
Gemma 3 1B Q4_K_M
1.0B
1.1 GB
Q4_K_M
32,768 ctx
Qwen42%
Qwen3 0.6B Q4_K_M
600M
1.2 GB
Q4_K_M
32,768 ctx
Qwen42%
Qwen3 0.6B FP16
600M
1.6 GB
FP16
32,768 ctx