Microsoft

🇺🇸United StatesWebsite

Top Model

67.4

avg score

Total Models

tracked on BenchGecko

Open Source

100% of models

$0.07

Cheapest Model

per 1M input tokens

50.2

Avg Benchmark

across 6 scored models

Model Categories

LLM7

$0.07

$0.62

Low: $0.07Median: $0.62High: $0.62

100%

7 open source0 proprietary

#▲	Model	Avg	ARC AI2?	BBH?	GSM8K?	HellaSwag?	LAMBADA?	MMLU?	GPQA diamond?	MATH level 5?	otis mock ?	WeirdML?	Winogrande?	SimpleBench?	aider poly?	lech mazur?	GSO-Bench?	fiction li?	swe bench ?	terminal b?	frontierma?	simpleqa v?	frontierma?	chess puzz?	APEX-Agents?	OSWorld?	ARC-AGI-2?	HLE?	TriviaQA?	ScienceQA?	PIQA?	OpenBookQA?	CadEval?	Balrog?	GeoBench?	Cybench?	ANLI?	the agent ?	VideoMME?	ARC-AGI?	deepresear?	VPCT?	$/1M in	Context	Released
1	phi-3-small 7.4B🇺🇸 MicrosoftOpen	67.4	87.6	72.1	-	69.3	-	67.6	-	-	-	-	63.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	58.1	-	-	84.0	-	-	-	-	37.1	-	-	-	-	-	-	-	Jan 242y ago
2	phi-3-mini 3.8B🇺🇸 MicrosoftOpen	61.0	79.9	62.3	-	68.9	-	58.4	-	-	-	-	41.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	64.0	-	-	84.0	-	-	-	-	29.2	-	-	-	-	-	-	-	Jan 242y ago
3	phi-3-medium 14B🇺🇸 MicrosoftOpen	58.6	88.8	75.2	-	76.5	-	70.7	3.5	17.6	-	-	63.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	73.9	-	-	83.2	-	-	-	-	33.7	-	-	-	-	-	-	-	Jan 242y ago
4	Phi 4🇺🇸 MicrosoftOpen	45.7	-	-	-	-	-	79.7	41.4	64.9	13.7	-	-	-	-	62.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	11.6	-	-	-	-	-	-	-	-	$0.07	16K	Jan 251y ago
5	Phi-2🇺🇸 MicrosoftOpen	41.2	67.9	45.9	-	38.1	-	44.5	-	-	-	-	9.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	45.2	-	-	64.8	-	-	-	-	13.8	-	-	-	-	-	-	-	Jan 242y ago
6	Phi-1.5🇺🇸 MicrosoftOpen	27.2	25.9	-	-	30.1	-	16.8	-	-	-	-	46.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	16.3	-	-	-	-	-	-	-	-	-	-	-	-	Jan 242y ago

90+ 80-89 70-79 60-69 <60Scores in % unless noted. Avg = unweighted mean across tested benchmarks.