Microsoft
Microsoft
Language model · United States ·Dec 2023

Phi-2

Language Open (Restricted)
2.7B
Parameters
10
Benchmarks

About Phi-2

Phi-2 is an AI model developed by Microsoft, in the language category, released in 2023, made available as an open-weights model with 2.7B params.

On this page you'll find Phi-2's full specifications, including 10 benchmark results. Review provider pricing and benchmark scores below, or compare Phi-2 head-to-head with other language models.

Links

Run Phi-2 locally — what to buy

Share this build ↗

Sized to what Phi-2 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.

Software support

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

Community

Would you run Phi-2 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-2 benchmark scores
Benchmark Category Score Variant
HumanEval code 47.6% cited-qwen2
MBPP code 55% cited-qwen2
MMLU general 52.7% cited-qwen2
ARC-Challenge knowledge 61.1 cited-qwen2
GSM8K math 57.2% cited-qwen2
MATH math 3.5% cited-qwen2
BIG-Bench Hard reasoning 43.4% cited-qwen2
HellaSwag reasoning 73.1 cited-qwen2
Winogrande reasoning 74.4 cited-qwen2
TruthfulQA safety 44.5% cited-qwen2

More models like Phi-2

Frequently asked questions

What is Phi-2?

Phi-2 is an AI model developed by Microsoft, in the language category, released in 2023. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created Phi-2?

Phi-2 was developed by Microsoft and released in 2023.

Is Phi-2 open source or proprietary?

Phi-2 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-2 cost?

Pricing for Phi-2 depends on the provider. See the providers table on this page for the latest API rates.

How does Phi-2 perform on benchmarks?

Phi-2 is benchmarked across 10 evaluations on GenAIList, including HumanEval (47.6%). 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-2?