Tested on 10 benchmarks · BenchGecko score 69.2. Top scores: ARC AI2 (88.8%), OpenBookQA (83.2%), HellaSwag (76.5%).
BIG-Bench Hard. 23 challenging tasks from BIG-Bench where prior language models fell below average human performance.
Competition-level math from AMC, AIME, and olympiad problems. Level 5 is the hardest tier, requiring creative problem-solving.
AI2 Reasoning Challenge. Grade-school science questions requiring multi-step reasoning. Easy and Challenge sets test different difficulty levels.
Elementary science questions with access to a small book of core science facts. Tests reasoning beyond memorization.
Sentence completion requiring commonsense reasoning about physical and social situations. Tests real-world understanding.
- Typetext
- ContextN/A
- ReleasedJan 2024
- LicenseOpen Source
- Statusbenchmark-only
Frequently Asked Questions
Key facts · as of 2026-04-09
- phi-3-medium 14B by Microsoft. BenchGecko score 69.2, rank 39 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-medium 14B · benchmarks, pricing and providers. BenchGecko, data as of 2026-04-09. https://benchgecko.ai/model/phi-3-medium-14b
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