Phi-3 Medium

Phi-3 Medium is a 14B-parameter Microsoft model with a 128K context window, delivering strong reasoning and coding performance at a size suitable for 12–16 GB VRAM consumer and prosumer GPUs.

Microsoft · Phi

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedJune 2026SourcesMicrosoft Phi-3 Medium model page

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Local assistant, reasoning, and long-context RAG workflows on 12–16 GB VRAM GPUs (RTX 3080/4070/4080). Good step up from 7B without requiring multi-GPU.

Who should use it

Local assistant, reasoning, and long-context RAG workflows on 12–16 GB VRAM GPUs (RTX 3080/4070/4080). Good step up from 7B without requiring multi-GPU.
Builders who want local or self-hosted testing options.

Common workflows

Mid-size local assistant and reasoning workflows
small workflows
edge workflows
local workflows
efficient workflows

Deployment and hardware notes

14B parameters. Q4_K_M requires approximately 8.4 GB VRAM — fits 10–12 GB consumer GPUs. Q8_0 requires approximately 15 GB VRAM. FP16 requires approximately 28 GB VRAM. 128K context supported via Ollama phi3:medium.

License and usage notes

MIT. Open weights. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

Open weights model option for Phi workflows.
Local assistant, reasoning, and long-context RAG workflows on 12–16 GB VRAM GPUs (RTX 3080/4070/4080). Good step up from 7B without requiring multi-GPU.
Use `ollama run phi3:medium` for the fastest local start. Q4 fits a 12 GB GPU; Q8 suits 16 GB or 24 GB setups.

Limitations

Requires 12+ GB VRAM for comfortable Q4 inference; tighter fit than 7B models on consumer hardware. Superseded for coding tasks by Phi-3.5-MoE and Qwen 2.5 Coder variants.
14B parameters. Q4_K_M requires approximately 8.4 GB VRAM — fits 10–12 GB consumer GPUs. Q8_0 requires approximately 15 GB VRAM. FP16 requires approximately 28 GB VRAM. 128K context supported via Ollama phi3:medium.
Context window and limits: 128K tokens.
Verify the exact model card, provider docs, license, and serving support before production use.