NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Phi 3 Mini 4k Instruct
Microsoft logo

Phi 3 Mini 4k Instruct

by Microsoft · Released Apr 2024

Open Source
50.8
avg score
Rank #141
Compare
Better than 55% of all models
Context
N/A
Input $/1M
n/a
Output $/1M
n/a
Type
text-generation
License
Open Source
Benchmarks
7 tested
Data as of
About

Microsoft text generation model. 744K downloads on HuggingFace.

Tested on 7 benchmarks · BenchGecko score 50.8. Top scores: Chatbot Arena Elo — Overall (1128.0%), IFEval (54.8%), BBH (HuggingFace) (36.6%).

Capabilities
reasoning
13.1
#160 globally
math
16.4
#246 globally
knowledge
22.3
#259 globally
language
54.8
#120 globally
general
36.6
#88 globally
Benchmark Scores
Compare All
Tested on 7 benchmarks · Ranked across 6 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
reasoningCompare reasoning →
MUSR

HuggingFace MuSR (Multi-Step Reasoning). Tests multi-hop reasoning requiring chaining multiple facts together.

13.1·
mathCompare math →
MATH Level 5

HuggingFace evaluation of MATH Level 5 problems. Competition math requiring advanced reasoning and proof construction.

16.4·
knowledgeCompare knowledge →
MMLU-PRO

HuggingFace MMLU-Pro. Harder version of MMLU with 10 answer choices instead of 4 and more challenging questions.

33.6·
GPQA

HuggingFace evaluation of GPQA (Graduate-Level Google-Proof Q&A). PhD-level science questions that cannot be easily searched.

11.0·
Quick compare:Google DeepMind logovs Gemini 2.0 Proanthracite-org logovs Magnum v4 72BGoogle DeepMind logovs Gemma 2 9B
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · Microsoft Phi 3
Apr 2024
50.8
n/an/actx7 benchmarks
Aug 2024
51.1+0.3
n/an/actx6 benchmarks
Aug 2024
n/an/actx
Jan 2024
69.2+18.1
n/an/actx10 benchmarks
Jan 2024
61.5-7.7
n/an/actx9 benchmarks
Jan 2024
79.2+17.7
n/an/actx8 benchmarks
See the full Phi 3 family →
Similar Models
Google DeepMind logo
Gemini 2.0 Pro
Google DeepMind
50.7n/a
anthracite-org logo
Magnum v4 72B
anthracite-org
50.9$3.00/1M
Google DeepMind logo
Gemma 2 9B
Google DeepMind
51.1$0.03/1M
Links
Info
MicrosoftPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@Microsoft
Source Code
GitHubHugging Face
BenchGecko API
microsoft-phi-3-mini-4k-instruct
Specifications
  • Typetext-generation
  • ContextN/A
  • ReleasedApr 2024
  • LicenseOpen Source
  • StatusActive
Available On
Microsoft logoMicrosoftn/a
Categories
reasoningmathknowledgelanguagegeneral
Learn More
transformeropen weightstokens
Share & Export
Tweet
Related Models
Google DeepMind logoGemini 2.0 Pro
50.7
anthracite-org logoMagnum v4 72B
50.9
Google DeepMind logoGemma 2 9B
51.1
Microsoft logoPhi 3.5 Mini Instruct
51.1
Phi 3 Mini 4k Instruct is an open-source text-generation AI model by Microsoft, released in April 2024. It has an average benchmark score of 50.8.

Related Models

Gemini 2.0 Pro · Google DeepMindMagnum v4 72B · anthracite-orgGemma 2 9B · Google DeepMindPhi 3.5 Mini Instruct · MicrosoftMiniMax M2.7 · minimax

Benchmarks

Chatbot Arena Elo — OverallIFEvalBBH (HuggingFace)MMLU-PROMATH Level 5

Related Pages

Microsoft · ProviderMicrosoft · EconomyAll ModelsCompare ModelsPricingDevelopers · API

Key facts · as of 2026-04-09

  • Phi 3 Mini 4k Instruct by Microsoft. BenchGecko score 50.8, rank 141 of 312 scored models (normalized average of public benchmark scores).
  • List price n/a input · n/a output per 1M tokens (as of 2026-04-09).

How to cite · data as of 2026-04-09

Phi 3 Mini 4k Instruct · benchmarks, pricing and providers. BenchGecko, data as of 2026-04-09. https://benchgecko.ai/model/microsoft-phi-3-mini-4k-instruct

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP