microsoft / text-generation updated 7 months ago

Phi-3.5-MoE-instruct

Name: Phi-3.5-MoE-instruct
Author: microsoft

Phi-3.5-MoE is a lightweight, state-of-the-art open model built upon datasets used for Phi-3 - synthetic data and filtered publicly available documents - with a focus on very high-quality, reasoning dense data. The model supports multilingual and comes with 128K context length (in tokens). The model underwent a rigorous enhancement process, incorporating supervised fine-tuning, proximal policy optimization, and direc...

Params

41.9 B

Context

131,072

Downloads 30d

161 K

Likes

574

Commercial use: allowed mit Not gated SAFETENSORS 1 languages View on Hugging Face ↗

Download history

daily snapshots · 10 days

161 K161 K

Jul 28Jul 31Aug 3Aug 6

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	bf16	83.7 GB	98.9 GB	❌ Won’t fit
model.safetensors (bf16, full)	bf16 + 128K ctx	83.7 GB	193.1 GB	❌ Won’t fit

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Run it

copy-paste, exact tags checked against the Hub

~ · curl · api/v1

$ curl -s https://aimodelscomparison.com/api/v1/models/phi-3-5-moe-instruct
{
  "hf_id": "microsoft/Phi-3.5-MoE-instruct",
  "params_b": 41.87,
  "context_length": 131072,
  "license": { "id": "mit", "commercial": "yes" },
  "downloads_30d": 161233,
  "vram_estimates": [
    { "quant": "bf16", "gb": 98.9 }
  ],
  "updated_at": "2026-07-28T18:08:58Z"
}

est. VRAM —on RTX 4090 · 24 GBJSON API →

Specifications

Architecture: PhiMoEForCausalLM
Parameters: 41.9 B
Tensor type: BF16
Context length: 131,072
Vocabulary: 32,064
Layers / heads: 32 / 32
Licence: mit
First seen on the Hub: 2024-08-17
Training datasets: undisclosed
Added to our catalog: 2026-07-28

Compare with

Qwen3.6-35B-A3B-NVFP4 11.3 M gpt-oss-20b 8.6 M Qwen3-32B 8.3 M dolphin-2.9.1-yi-1.5-34b 4.6 M OTel-2.0-LLM-31B-IT 3.3 M Qwen3-30B-A3B 3.1 M

Appears in

Best open coding models Commercial-use text models Models that fit in 12 GB