Noometry

Model comparison

DeepSeek-V3 vs GPT-5.2

GPT-5.2 is the stronger model overall, scoring 54.1 to 39.5 on the Noometry Index. DeepSeek-V3 costs 12× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.

Last verified . 30 shared benchmarks.

DeepSeek-V3 DeepSeek

39.5

Rank #166 Confirmed

GPT-5.2 OpenAI

54.1

Rank #34 Confirmed

Summary

  • They share 30 benchmarks with published results for both. DeepSeek-V3 scores higher in 0 categories and GPT-5.2 in 8 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where GPT-5.2 leads 50.2 to 20.5.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 96.1% for GPT-5.2.
  • DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
  • GPT-5.2 accepts more context: 400K tokens versus 164K.
  • DeepSeek-V3 has downloadable open weights; the other is API-only.

Side by side

DeepSeek-V3GPT-5.2
ProviderDeepSeekOpenAI
Noometry Index39.554.1
Released2024-12-262025-12-11
WeightsOpenProprietary
Context window164K400K
Max output164K128K
Input $ / M tokens$0.24$1.75
Output $ / M tokens$0.90$14
Results tracked6067

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

Category by category

Coding GPT-5.2 leads

DeepSeek-V3: 42.3 (#106), GPT-5.2: 51.6 (#37)

BenchmarkDeepSeek-V3GPT-5.2
WeirdML36.1%72.2%
LMArena Coding13681447
SWE-bench Verified—73.8%
SWE-bench Verified (bash only)—72.8%
Aider Polyglot55.1%—
LMArena WebDev—1416
SWE-bench Multilingual—66.7%
SciCode35.8%—
GSO—27.4%
BigCodeBench Instruct50%—
LiveBench Coding70.9%—
BigCodeBench Complete62.2%—
ALE-Bench—1,294
AlgoTune—2.05
HumanEval+86.6%—
MBPP+73%—

Agentic & Tool Use Not comparable

DeepSeek-V3: —, GPT-5.2: 40.2 (#24)

BenchmarkDeepSeek-V3GPT-5.2
METR Time Horizons49.6%75.3%
Terminal-Bench—64.9%
Berkeley Function Calling Leaderboard—55.9%
GDPval—49.7%
Remote Labor Index—2.5%
τ²-bench Airline—83%
τ²-bench Banking—32.2%
τ²-bench Retail—81.6%
τ²-bench Telecom—89.7%
DeepResearch Bench—41.1%
LMArena Search—1207
Vending-Bench 2—3,591

Reasoning GPT-5.2 leads

DeepSeek-V3: 20.5 (#236), GPT-5.2: 50.2 (#35)

BenchmarkDeepSeek-V3GPT-5.2
SimpleBench27.2%45.8%
Kagi LLM Benchmark52.3%73.3%
LMArena Hard Prompts13651445
DTBench64.8%90.9%
LMCA15.5%43.9%
Epoch Capabilities Index135.94153.45
ForecastBench59.160.1
ARC-AGI-2—52.9%
NYT Connections (extended)—83.6%
ARC-AGI-1—86.2%
CritPt0%—
Chess Puzzles—49%
EnigmaEval—10.4%
EBR-Bench—23%
LiveBench Reasoning65.8%—
Mystery Game Puzzles—23%
LiveBench Data Analysis60.9%—
BIG-Bench Hard87.5%—
HellaSwag88.9%—
LiveBench66.9%—
PIQA84.7%—
WinoGrande85.2%—

Math GPT-5.2 leads

DeepSeek-V3: 32.1 (#219), GPT-5.2: 60.0 (#38)

BenchmarkDeepSeek-V3GPT-5.2
OTIS Mock AIME 2024-202537.8%96.1%
LMArena Math13731440
FrontierMath (Feb 2025 set)1.7%40.7%
FrontierMath (Tiers 1-3)—67.4%
FrontierMath Tier 4—31.7%
MathArena Final-Answer Competitions—72%
ProofBench—15%
Omni-MATH40.3%—
LiveBench Math73.5%—
MATH Level 575.5%—
FrontierMath Tier 4 (v1)—18.8%

Knowledge GPT-5.2 leads

DeepSeek-V3: 37.5 (#155), GPT-5.2: 59.3 (#32)

BenchmarkDeepSeek-V3GPT-5.2
GPQA Diamond67.6%91.4%
Vectara Hallucination Rate6.1%8.4%
LMArena Expert13511445
Humanity's Last Exam—27.8%
SimpleQA Verified—37.1%
MMLU-Pro72.3%—
Confabulations26.1%—
GPQA (HELM)53.8%—
ARC (AI2) Challenge95.3%—
MMLU87.2%—
TriviaQA82.9%—

Multimodal Not comparable

DeepSeek-V3: —, GPT-5.2: 51.3 (#7)

BenchmarkDeepSeek-V3GPT-5.2
LMArena Vision—1268
VPCT—84%
Furniture Assembly—38.3%
LMArena Document—1405

Multilingual GPT-5.2 leads

DeepSeek-V3: 48.5 (#143), GPT-5.2: 53.4 (#67)

BenchmarkDeepSeek-V3GPT-5.2
LMArena Non-English13581425
LMArena Chinese13911460
LMArena French13851455
LMArena German13741448
LMArena Japanese13331420
LMArena Korean13191392
LMArena Russian13731440
LMArena Spanish13581433

Instruction Following GPT-5.2 leads

DeepSeek-V3: 72.8 (#130), GPT-5.2: 74.7 (#89)

BenchmarkDeepSeek-V3GPT-5.2
LMArena Instruction Following13451417
LiveBench Instruction Following81.5%—
IFEval83.2%—

Long Context GPT-5.2 leads

DeepSeek-V3: 34.0 (#253), GPT-5.2: 44.0 (#78)

BenchmarkDeepSeek-V3GPT-5.2
LMArena Longer Query13521428
Fiction.LiveBench50%—
CL-bench—18.2%

Writing & Preference GPT-5.2 leads

DeepSeek-V3: 57.4 (#130), GPT-5.2: 66.8 (#32)

BenchmarkDeepSeek-V3GPT-5.2
LMArena Text13751439
LMArena Creative Writing13641401
EQ-Bench Creative Writing14721703
LMArena Multi-Turn13891458
Short-Story Creative Writing77%—
WildBench83%—
LiveBench Language49.1%—

Frequently asked questions

Is DeepSeek-V3 better than GPT-5.2?

GPT-5.2 is the stronger model overall, scoring 54.1 to 39.5 on the Noometry Index. DeepSeek-V3 costs 12× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.

Which is cheaper, DeepSeek-V3 or GPT-5.2?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; GPT-5.2 lists at $1.75 and $14.

Is DeepSeek-V3 or GPT-5.2 better for coding?

GPT-5.2 scores higher on coding benchmarks: 51.6 versus 42.3 in the Noometry coding category.

Which has the bigger context window?

GPT-5.2 does, with 400K tokens against 164K.

How many benchmarks do DeepSeek-V3 and GPT-5.2 share?

30 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and GPT-5.2 has 67.

Related comparisons

Go deeper