Model comparison

DeepSeek-V3 vs GPT-5 Mini

GPT-5 Mini is the stronger model overall, scoring 41.8 to 39.5 on the Noometry Index. DeepSeek-V3 costs 1.7× less per token, which makes it the better buy when GPT-5 Mini's lead doesn't matter for your workload.

Last verified October 10, 2026. 39 shared benchmarks.

DeepSeek-V3

DeepSeek

39.5

Rank #166 Confirmed

GPT-5 Mini OpenAI

41.8

Rank #128 Confirmed

Summary

They share 39 benchmarks with published results for both. DeepSeek-V3 scores higher in 2 categories and GPT-5 Mini in 6 categories; 7 gaps are clear of the uncertainty.
The widest gap is in math, where GPT-5 Mini leads 46.7 to 32.1.
The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 86.7% for GPT-5 Mini.
DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $0.25 / $2 for GPT-5 Mini.
GPT-5 Mini accepts more context: 400K tokens versus 164K.
DeepSeek-V3 has downloadable open weights; the other is API-only.

Side by side

DeepSeek-V3 and GPT-5 Mini specifications
	DeepSeek-V3	GPT-5 Mini
Provider	DeepSeek	OpenAI
Noometry Index	39.5	41.8
Released	2024-12-26	2025-08-07
Weights	Open	Proprietary
Context window	164K	400K
Max output	164K	128K
Input $ / M tokens	$0.24	$0.25
Output $ / M tokens	$0.90	$2
Results tracked	60	60

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

Category by category

Coding DeepSeek-V3 leads

DeepSeek-V3: 42.3 (#106), GPT-5 Mini: 40.1 (#146)

Coding benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
SciCode	35.8%	39.2%
WeirdML	36.1%	52.7%
LMArena Coding	1368	1406
SWE-bench Verified	—	64.7%
SWE-bench Verified (bash only)	—	59.8%
Aider Polyglot	55.1%	—
SWE-bench Multilingual	—	39.7%
BigCodeBench Instruct	50%	—
LiveBench Coding	70.9%	—
BigCodeBench Complete	62.2%	—
ALE-Bench	—	799.77
AlgoTune	—	1.38
HumanEval+	86.6%	—
MBPP+	73%	—

Agentic & Tool Use Not comparable

DeepSeek-V3: —, GPT-5 Mini: 31.1 (#70)

Agentic & Tool Use benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
Terminal-Bench	—	34.8%
Berkeley Function Calling Leaderboard	—	55.5%
METR Time Horizons	49.6%	—
Vending-Bench 2	—	-31.18

Reasoning GPT-5 Mini leads

DeepSeek-V3: 20.5 (#236), GPT-5 Mini: 23.9 (#168)

Reasoning benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
Kagi LLM Benchmark	52.3%	70.3%
CritPt	0%	0%
LMArena Hard Prompts	1365	1380
DTBench	64.8%	80.5%
LMCA	15.5%	34.2%
Epoch Capabilities Index	135.94	145.52
ForecastBench	59.1	61
ARC-AGI-2	—	4.4%
SimpleBench	27.2%	—
ARC-AGI-1	—	54.3%
Chess Puzzles	—	30%
EnigmaEval	—	8.2%
LiveBench Reasoning	65.8%	—
Mystery Game Puzzles	—	10%
LiveBench Data Analysis	60.9%	—
BIG-Bench Hard	87.5%	—
HellaSwag	88.9%	—
LiveBench	66.9%	—
PIQA	84.7%	—
WinoGrande	85.2%	—

Math GPT-5 Mini leads

DeepSeek-V3: 32.1 (#219), GPT-5 Mini: 46.7 (#69)

Math benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
OTIS Mock AIME 2024-2025	37.8%	86.7%
Omni-MATH	40.3%	72.2%
LMArena Math	1373	1378
MATH Level 5	75.5%	97.8%
FrontierMath (Feb 2025 set)	1.7%	27.2%
FrontierMath (Tiers 1-3)	—	46.7%
FrontierMath Tier 4	—	12.2%
ProofBench	—	9%
LiveBench Math	73.5%	—
FrontierMath Tier 4 (v1)	—	6.3%

Knowledge GPT-5 Mini leads

DeepSeek-V3: 37.5 (#155), GPT-5 Mini: 45.6 (#86)

Knowledge benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
GPQA Diamond	67.6%	75%
MMLU-Pro	72.3%	83.5%
Confabulations	26.1%	13.3%
Vectara Hallucination Rate	6.1%	12.9%
GPQA (HELM)	53.8%	75.6%
LMArena Expert	1351	1379
Humanity's Last Exam	—	19.4%
SimpleQA Verified	—	21.6%
ARC (AI2) Challenge	95.3%	—
MMLU	87.2%	—
TriviaQA	82.9%	—

Multimodal Not comparable

DeepSeek-V3: —, GPT-5 Mini: 35.6 (#85)

Multimodal benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
LMArena Vision	—	1202
VPCT	—	40.2%

Multilingual Too close to call

DeepSeek-V3: 48.5 (#143), GPT-5 Mini: 48.9 (#137)

Multilingual benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
LMArena Non-English	1358	1363
LMArena Chinese	1391	1385
LMArena French	1385	1386
LMArena German	1374	1366
LMArena Japanese	1333	1341
LMArena Korean	1319	1308
LMArena Russian	1373	1362
LMArena Spanish	1358	1355

Instruction Following GPT-5 Mini leads

DeepSeek-V3: 72.8 (#130), GPT-5 Mini: 76.2 (#46)

Instruction Following benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
IFEval	83.2%	92.7%
LMArena Instruction Following	1345	1357
LiveBench Instruction Following	81.5%	—

Long Context GPT-5 Mini leads

DeepSeek-V3: 34.0 (#253), GPT-5 Mini: 41.9 (#132)

Long Context benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
Fiction.LiveBench	50%	69.4%
LMArena Longer Query	1352	1355

Writing & Preference DeepSeek-V3 leads

DeepSeek-V3: 57.4 (#130), GPT-5 Mini: 55.2 (#148)

Writing & Preference benchmarks
Benchmark	DeepSeek-V3	GPT-5 Mini
LMArena Text	1375	1373
LMArena Creative Writing	1364	1325
Short-Story Creative Writing	77%	83.1%
EQ-Bench Creative Writing	1472	1313
WildBench	83%	85.5%
LMArena Multi-Turn	1389	1363
LiveBench Language	49.1%	—

Frequently asked questions

Is DeepSeek-V3 better than GPT-5 Mini?

Which is cheaper, DeepSeek-V3 or GPT-5 Mini?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; GPT-5 Mini lists at $0.25 and $2.

Is DeepSeek-V3 or GPT-5 Mini better for coding?

DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 40.1 in the Noometry coding category.

Which has the bigger context window?

GPT-5 Mini does, with 400K tokens against 164K.

How many benchmarks do DeepSeek-V3 and GPT-5 Mini share?

39 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and GPT-5 Mini has 60.