Model comparison

DeepSeek-V3 vs Phi-4

DeepSeek-V3 is the stronger model overall, scoring 39.2 to 30.9 on the Noometry Index. Phi-4 costs 4.6× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.

Last verified October 10, 2026. 34 shared benchmarks.

DeepSeek-V3

DeepSeek

39.2

Rank #157 Confirmed

Phi-4

Microsoft

30.9

Rank #269 Confirmed

Summary

They share 34 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Phi-4 in 1 category; 8 gaps are clear of the uncertainty.
The widest gap is in writing & preference, where DeepSeek-V3 leads 57.6 to 40.7.
The biggest single-benchmark swing is LiveBench Coding: 70.9% for DeepSeek-V3 and 30.7% for Phi-4.
Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.24 / $0.90 for DeepSeek-V3.
DeepSeek-V3 accepts more context: 164K tokens versus 128K.

Side by side

DeepSeek-V3 and Phi-4 specifications
	DeepSeek-V3	Phi-4
Provider	DeepSeek	Microsoft
Noometry Index	39.2	30.9
Released	2024-12-26	2024-12-11
Weights	Open	Open
Context window	164K	128K
Max output	164K	4K
Input $ / M tokens	$0.24	$0.07
Output $ / M tokens	$0.90	$0.14
Results tracked	60	37

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding DeepSeek-V3 leads

DeepSeek-V3: 41.9 (#99), Phi-4: 33.8 (#237)

Coding benchmarks
Benchmark	DeepSeek-V3	Phi-4
BigCodeBench Instruct	50%	45.5%
LiveBench Coding	70.9%	30.7%
LMArena Coding	1368	1231
BigCodeBench Complete	62.2%	55.4%
Aider Polyglot	55.1%	—
SciCode	35.8%	—
WeirdML	36.1%	—
HumanEval+	86.6%	—
MBPP+	73%	—

Agentic & Tool Use Not comparable

DeepSeek-V3: —, Phi-4: 22.6 (#120)

Agentic & Tool Use benchmarks
Benchmark	DeepSeek-V3	Phi-4
Berkeley Function Calling Leaderboard	—	28.8%
BALROG	—	11.6%
METR Time Horizons	49.6%	—

Reasoning DeepSeek-V3 leads

DeepSeek-V3: 20.0 (#226), Phi-4: 17.0 (#287)

Reasoning benchmarks
Benchmark	DeepSeek-V3	Phi-4
LiveBench Reasoning	65.8%	47.8%
LMArena Hard Prompts	1365	1220
LiveBench Data Analysis	60.9%	45.2%
Epoch Capabilities Index	135.95	130.43
LiveBench	66.9%	41.6%
SimpleBench	27.2%	—
Kagi LLM Benchmark	52.3%	—
CritPt	0%	—
Chess Puzzles	—	1%
DTBench	64.8%	—
LMCA	15.5%	—
BIG-Bench Hard	87.5%	—
ForecastBench	59.1	—
HellaSwag	88.9%	—
PIQA	84.7%	—
WinoGrande	85.2%	—

Math DeepSeek-V3 leads

DeepSeek-V3: 30.4 (#208), Phi-4: 19.6 (#279)

Math benchmarks
Benchmark	DeepSeek-V3	Phi-4
OTIS Mock AIME 2024-2025	37.8%	13.8%
LiveBench Math	73.5%	42%
LMArena Math	1373	1246
MATH Level 5	75.5%	64.9%
Omni-MATH	40.3%	—
FrontierMath (Feb 2025 set)	1.7%	—

Knowledge DeepSeek-V3 leads

DeepSeek-V3: 37.5 (#147), Phi-4: 32.5 (#204)

Knowledge benchmarks
Benchmark	DeepSeek-V3	Phi-4
GPQA Diamond	67.6%	56.1%
Confabulations	26.1%	29.4%
Vectara Hallucination Rate	6.1%	3.7%
LMArena Expert	1352	1203
MMLU	87.2%	84.8%
MMLU-Pro	72.3%	—
GPQA (HELM)	53.8%	—
ARC (AI2) Challenge	95.3%	—
TriviaQA	82.9%	—

Multilingual DeepSeek-V3 leads

DeepSeek-V3: 48.8 (#138), Phi-4: 37.4 (#232)

Multilingual benchmarks
Benchmark	DeepSeek-V3	Phi-4
LMArena Non-English	1358	1197
LMArena Chinese	1390	1211
LMArena French	1386	1223
LMArena German	1374	1222
LMArena Japanese	1333	1158
LMArena Korean	1321	1151
LMArena Russian	1372	1209
LMArena Spanish	1358	1234

Instruction Following DeepSeek-V3 leads

DeepSeek-V3: 73.0 (#125), Phi-4: 60.6 (#246)

Instruction Following benchmarks
Benchmark	DeepSeek-V3	Phi-4
LiveBench Instruction Following	81.5%	58.4%
LMArena Instruction Following	1345	1201
IFEval	83.2%	—

Long Context Phi-4 leads

DeepSeek-V3: 34.0 (#248), Phi-4: 36.9 (#221)

Long Context benchmarks
Benchmark	DeepSeek-V3	Phi-4
LMArena Longer Query	1353	1217
Fiction.LiveBench	50%	—

Writing & Preference DeepSeek-V3 leads

DeepSeek-V3: 57.6 (#124), Phi-4: 40.7 (#239)

Writing & Preference benchmarks
Benchmark	DeepSeek-V3	Phi-4
LMArena Text	1375	1217
LMArena Creative Writing	1364	1182
Short-Story Creative Writing	77%	62.6%
LMArena Multi-Turn	1389	1206
LiveBench Language	49.1%	25.6%
EQ-Bench Creative Writing	1472	—
WildBench	83%	—

Frequently asked questions

Is DeepSeek-V3 better than Phi-4?

Which is cheaper, DeepSeek-V3 or Phi-4?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; DeepSeek-V3 lists at $0.24 and $0.90.

Is DeepSeek-V3 or Phi-4 better for coding?

DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 33.8 in the Noometry coding category.

Which has the bigger context window?

DeepSeek-V3 does, with 164K tokens against 128K.

How many benchmarks do DeepSeek-V3 and Phi-4 share?

34 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Phi-4 has 37.