Phi-3.5 models
4 Phi-3.5 models on Hugging Face, from 3.8B to 41.9B parameters. At the precision each one is published in, the smallest needs about 8.5 GB of VRAM (Phi-3.5-mini-instruct, cheapest live fit: RTX 5060 Ti) and the largest about 93.6 GB (Phi-3.5-MoE-instruct, cheapest live fit: RTX PRO 6000). The cheapest way to run Phi-3.5-mini-instruct is $0.110/hr.
Phi-3.5 models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Phi-3.5-mini-instruct | 3.8B | BF16 | 8.5 GB | RTX 5060 Ti | $0.110/hr |
| Phi-3.5-vision-instruct | 4.1B | BF16 | 9.3 GB | RTX 5060 Ti | $0.110/hr |
| VLM2Vec-Full | 4.1B | BF16 | 9.3 GB | RTX 5060 Ti | $0.110/hr |
| Phi-3.5-MoE-instruct | 41.9B | BF16 | 93.6 GB | RTX PRO 6000 | $1.38/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; see the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
More from Microsoft
- Florence-2 (5 models)
- Phi-4 (7 models)
- Phi-3 (6 models)
- VibeVoice (4 models)
- microsoft (6 models)
