Microsoft
Microsoft
Language model · United States ·Apr 2024

phi-3.5-mini

Language Open (Restricted)
3.8B
Parameters
13
Benchmarks

About phi-3.5-mini

phi-3.5-mini is an AI model developed by Microsoft, in the language category, released in 2024, made available as an open-weights model with 3.8B params.

On this page you'll find phi-3.5-mini's full specifications, including 13 benchmark results. Review provider pricing and benchmark scores below, or compare phi-3.5-mini head-to-head with other language models.

→ More models by Microsoft → All language models → AI benchmarks leaderboard

Links

Announcement ↗ Hugging Face ↗

Run phi-3.5-mini locally — what to buy

Share this build ↗

Sized to what phi-3.5-mini actually needs (~8k context at Q4), not to the biggest GPU: Good = cheapest that runs it, Better = best value, Best = most headroom. Speed is a hardware estimate; anything past ~120 tok/s is shown as instant.

Best
Intel Arc B580 12GB
Consumer GPU · ~$249
~120 tok/s
Q4 · fits full context · up to 8k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →

Software support

·ExLlamaV2 ·KoboldCpp ·LM Studio ·MLX ·Ollama ·SGLang ·TensorRT-LLM ·llama.cpp ·vLLM ?KTransformers ?text-generation-webui

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Consumer GPU
75.8–120+ tok/s
32 options →
Apple
45.8–120+ tok/s
18 options →
Datacenter GPU
120+ tok/s
13 options →

Community

Would you run phi-3.5-mini again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

phi-3.5-mini benchmark scores▶
Compare on the benchmark table →
Benchmark Category Score Variant
ArenaHard general 34.4% cited-phi4-mini
MMLU general 65.5% cited-phi4-mini
MMLU-Pro general 47.4% cited-phi4-mini
MMMLU general 51.8% cited-phi4-mini
ARC-Challenge knowledge 84.6 cited-phi4-mini
GSM8K math 76.9% cited-phi4-mini
MATH math 49.8% cited-phi4-mini
MGSM math 49.6 cited-phi4-mini
BIG-Bench Hard reasoning 63.1% cited-phi4-mini
HellaSwag reasoning 72.2 cited-phi4-mini
PIQA reasoning 78.2 cited-phi4-mini
Winogrande reasoning 72.2 cited-phi4-mini
TruthfulQA safety 65.2% cited-phi4-mini

More models like phi-3.5-mini

AI
DMind-1
 AI
COLLECTIVE-1
 AI
Cogito 1 series
 AI
Euler (Euler大模型)
· 2.5B
 AI
SmolLM-1.7B
· 1.7B
 AI
Llama-3-70B-Synthia-v3.5
· 70B

Frequently asked questions

What is phi-3.5-mini? ▶

phi-3.5-mini is an AI model developed by Microsoft, in the language category, released in 2024. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created phi-3.5-mini? ▶

phi-3.5-mini was developed by Microsoft and released in 2024.

Is phi-3.5-mini open source or proprietary? ▶

phi-3.5-mini ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

How much does phi-3.5-mini cost? ▶

Pricing for phi-3.5-mini depends on the provider. See the providers table on this page for the latest API rates.

How does phi-3.5-mini perform on benchmarks? ▶

phi-3.5-mini is benchmarked across 13 evaluations on GenAIList, including ArenaHard (34.4%). See the full benchmark table below and compare it with other models.

No videos yet. Know a good one? Add it below.

Know a great video about phi-3.5-mini?