BestLLMfor EN Your hardware. Your LLM. Your call.
APIOpen data Find my LLM
Model fiche

Phi-4 Mini Reasoning 3.8B

By Microsoft · United States

Updated 2026-07-13

reasoning small
Parameters
3.8B
License
MIT
Context
125k
VRAM (Q4)
10 GB
Released
April 2025

Overview

Microsoft's 3.8B Phi-4 Mini variant trained on R1-style reasoning traces under MIT. AIME24 57.5 and MATH-500 94.6 — remarkable math chops for the size.

When to pick this model

  • Math and STEM reasoning on a laptop
  • Educational tutoring apps under MIT
  • Research into small-model reasoning distillation
  • Battery-constrained reasoning workloads

VRAM requirements by quantization

VRAM REQUIRED (GB)812162432Q4_K_M10 GBQ5_K_M12 GBQ8_018 GBFP1633 GB
QuantizationVRAM required
Q4_K_M (recommended)10 GB
Q5_K_M12 GB
Q8_018 GB
FP16 (no quantization)33 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

In practice, Phi-4 Mini Reasoning 3.8B needs a 12 GB card at Q4_K_M (10 GB). Stepping up to Q8_0 nearly doubles the footprint to 18 GB, and unquantized FP16 weights take 33 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.

Without a GPU, Phi-4 Mini Reasoning 3.8B needs roughly 12 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 14 tokens/sec on entry-level GPUs, on the order of 40 tokens/sec on a mid-range card, and up to 100 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.

What hardware do you need

The table below matches Phi-4 Mini Reasoning 3.8B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.

GPU memoryExample cardsBest fit for Phi-4 Mini Reasoning 3.8B
8 GBRTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GBDoes not fit — needs 10 GB at Q4_K_M
12 GBRTX 5070, RTX 5070 Ti Laptop, RTX 4080 LaptopQ5_K_M (12 GB used)
16 GBRTX 5080, RTX 4080 Super, Radeon RX 9070 XTQ5_K_M (12 GB used)
24 GBRTX 4090, Radeon RX 7900 XTX, RTX 5090 LaptopQ8_0 (18 GB used)
32 GBRTX 5090Q8_0 (18 GB used)

Which GPU should you buy to run Phi-4 Mini Reasoning 3.8B?

To run Phi-4 Mini Reasoning 3.8B locally at Q4, you need ~10 GB of VRAM. The best value for this is a RTX 5070 (12 GB VRAM).

Check RTX 5070 price on Amazon →

As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.

Published benchmark scores

BenchmarkScore
AIME 202457.5
MATH-50094.6

Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.

To put Phi-4 Mini Reasoning 3.8B in context: its AIME 2024 score of 57.5 ranks #7 of the 8 catalog models with a published AIME 2024 result (catalog median 71.7); its MATH-500 score of 94.6 ranks #2 of the 8 catalog models with a published MATH-500 result (catalog median 93.3). Scores are author-published and measured under different harnesses, so treat gaps of a few points as noise rather than a verdict.

Strengths

  • AIME24 57.5 — exceptional for 3.8B
  • MATH-500 94.6 nearly matches frontier models
  • Fits comfortably on any laptop
  • MIT license

Limitations

  • English-first
  • Verbose CoT typical of reasoning models
  • Outside math, quality trails the base Phi-4 Mini

Typical workloads

In our catalog grid, Phi-4 Mini Reasoning 3.8B is filed under On-device Math, Edge Logic — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: multi-step reasoning and math-flavoured tasks.

The 125k-token context window covers long chats and mid-sized documents, though very large retrieval workloads will need chunking. The MIT license is permissive, so shipping it inside a commercial product raises no special legal questions.

Architecture & training

Architecture: Dense 3.8B · trained on R1 traces

Training: Synthetic reasoning distillation.

Verdict

Pound-for-pound the most impressive small reasoner under MIT — pick it for math on the smallest hardware.

Quick start

ollama run phi4-mini-reasoning:3.8b

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.

Similar models worth comparing

Frequently asked questions

How much VRAM does Phi-4 Mini Reasoning 3.8B need?

At the recommended Q4_K_M quantization, Phi-4 Mini Reasoning 3.8B needs about 10 GB of VRAM. Q8_0 takes 18 GB, and unquantized FP16 weights take 33 GB.

Can Phi-4 Mini Reasoning 3.8B run without a GPU?

Yes — with roughly 12 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.

What context window does Phi-4 Mini Reasoning 3.8B support?

Phi-4 Mini Reasoning 3.8B supports a 125k-token context window (128,000 tokens).

Can I use Phi-4 Mini Reasoning 3.8B commercially?

Yes. Phi-4 Mini Reasoning 3.8B is released under MIT, a permissive open-source license that allows commercial use, modification and redistribution.

How fast is Phi-4 Mini Reasoning 3.8B on consumer hardware?

Our compatibility engine estimates on the order of 40 tokens/sec on a mid-range GPU and up to 100 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.

Which quantization of Phi-4 Mini Reasoning 3.8B should I download first?

Start with Q4_K_M (10 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. On a 24 GB card you can run up to Q8_0.

Tools

Is Phi-4 Mini Reasoning 3.8B the right pick for you?

Compute self-hosted ROI → Back to catalog