What GPU do I need to run microsoft/phi-1_5?
1.4B parameters, published in F16. View on Hugging Face
Set up phi-1_5phi-1_5 is published by microsoft on Hugging Face, with 55,389 downloads and 1,362 likes to date. It's a PhiForCausalLM model built for text-generation, published natively in F16.
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.
| Precision | Weight size | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| FP16 | 2.6 GB | 3.2 GB | V100 | 1 | $0.088/hr |
| FP8 (quantized) | 1.3 GB | 1.6 GB | RTX 4000 SFF Ada | 1 | $0.198/hr |
| INT4 (quantized) | 0.7 GB | 0.8 GB | RTX 3060 | 1 | $0.110/hr |
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run phi-1_5 at its published (F16) precision: 1× V100, at $0.088/hr per GPU ($0.088/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
phi-1_5: KV cache by context length
The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.
Multi-head attention: every layer caches keys and values for every head. Cached per token: 196,608 bytes at 16-bit.
| Context | KV cache, one sequence |
|---|---|
| 2K (model maximum) | 0.38 GB |
Computed from the layer, head and window counts in the model's own configuration (microsoft/phi-1_5), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
phi-1_5: common questions
How much VRAM does phi-1_5 need?
3.2 GB at FP16, 1.6 GB at FP8 (quantized), 0.8 GB at INT4 (quantized). All of those sit under 6 GB, the smallest capacity this page reasons about, so VRAM is not what limits where this model runs. The 2.6 GB of weights plus inference overhead is the whole requirement.
How many copies of phi-1_5 fit on one V100?
5, by VRAM alone. That card carries 16.0 GB and one copy needs 3.2 GB at FP16, on a live rate of $0.088/hr for the whole card. Throughput is not modelled here, so 5 copies is not 5 times the requests served.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Phi-1 models
- phi-1 (1.4B, F16)
Alternatives at this size
Other models for text-generation within about a third of phi-1_5's 1.4B parameters, from other model lines.
- Qwen2.5-1.5B-Instruct (1.5B, BF16)
- Llama-3.2-1B-Instruct (1.2B, BF16)
- gemma-3-1b-it (1000M, BF16)
- OpenELM-1_1B-Instruct (1.1B, BF16)
- OLMo-2-0425-1B (1.5B, F32)
More on phi-1_5
Fits on an 8 GB GPU at FP16: every model that fits in 8 GB.
Best chat and assistants models: how phi-1_5 ranks against the rest.
Related reading: V100 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, and Best GPU for LLM inference.
