Can the NVIDIA RTX 4090 run Phi-3.5 Mini 3.8B?
Yes
QuantFP16Full quality
Memory11.5 GBof 24 GB
Speed~79tokens/s
Yes. The NVIDIA RTX 4090 can run Phi-3.5 Mini 3.8B fully in VRAM at FP16 (full quality), using about 11.5 GB of the 24 GB available, at roughly 79 tokens/second.
Try other context lengths in the calculator →Memory needed by quantization
At 8,192 tokens of context.
| Quant | Quality | Memory | Fits? | Speed |
|---|---|---|---|---|
| FP16 | Full quality | 11.5 GB | Fits | ~79 tok/s |
| Q8_0 | Near-lossless | 7.9 GB | Fits | ~149 tok/s |
| Q6_K | Excellent | 7 GB | Fits | ~193 tok/s |
| Q5_K_M | Very good | 6.5 GB | Fits | ~223 tok/s |
| Q4_K_M | Good, most popular | 6.1 GB | Fits | ~261 tok/s |
| Q3_K_M | Noticeably degraded | 5.7 GB | Fits | ~324 tok/s |
| Q2_K | Heavily degraded | 5.3 GB | Fits | ~422 tok/s |
How to run Phi-3.5 Mini 3.8B on the NVIDIA RTX 4090
With Ollama installed, one command downloads and starts it:
ollama run phi3.5:3.8bOllama's default download is Q4_K_M (about 6.1 GB here). You have room for FP16, which is full quality: pick that GGUF in LM Studio or from Hugging Face.
GPUs that run Phi-3.5 Mini 3.8B comfortably
- NVIDIA RTX 4060 (8 GB)
- NVIDIA RTX 3070 (8 GB)
- NVIDIA RTX 3080 10GB
- NVIDIA RTX 3060 12GB
- NVIDIA RTX 4070 (12 GB)
Models that run well on the NVIDIA RTX 4090
Run it in the cloud
Don't want to buy hardware? Run Phi-3.5 Mini 3.8B on a rented GPU today.
Rent a cloud GPU by the hour. No upfront cost, cancel any time.