Model comparison

Claude Opus 4.5 vs DeepSeek-V3.2-Exp

Claude Opus 4.5 is the stronger model overall, scoring 50.5 to 44.3 on the Noometry Index. DeepSeek-V3.2-Exp costs 34× less per token, which makes it the better buy when Claude Opus 4.5's lead doesn't matter for your workload.

Last verified October 12, 2026. 40 shared benchmarks.

Claude Opus 4.5 Anthropic

50.5

Rank #47 Confirmed

DeepSeek-V3.2-Exp

DeepSeek

44.3

Rank #78 Confirmed

Summary

They share 40 benchmarks with published results for both. Claude Opus 4.5 scores higher in 7 categories and DeepSeek-V3.2-Exp in 2 categories; 9 gaps are clear of the uncertainty.
The widest gap is in reasoning, where Claude Opus 4.5 leads 42.6 to 22.1.
The biggest single-benchmark swing is ARC-AGI-2: 37.6% for Claude Opus 4.5 and 4% for DeepSeek-V3.2-Exp.
DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $5 / $25 for Claude Opus 4.5.
Claude Opus 4.5 accepts more context: 200K tokens versus 164K.
DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.

Side by side

Claude Opus 4.5 and DeepSeek-V3.2-Exp specifications
	Claude Opus 4.5	DeepSeek-V3.2-Exp
Provider	Anthropic	DeepSeek
Noometry Index	50.5	44.3
Released	2025-11-01	2025-09-29
Weights	Proprietary	Open
Context window	200K	164K
Max output	64K	66K
Input $ / M tokens	$5	$0.26
Output $ / M tokens	$25	$0.38
Results tracked	69	49

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

Category by category

Coding Claude Opus 4.5 leads

Claude Opus 4.5: 54.8 (#27), DeepSeek-V3.2-Exp: 46.5 (#65)

Coding benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
SWE-bench Verified (bash only)	76.8%	70%
LMArena WebDev	1494	1362
SWE-bench Multilingual	70.7%	59%
WeirdML	63.7%	39.5%
LMArena Coding	1504	1454
SWE-bench Verified	76.7%	—
Aider Polyglot	—	74.2%
SciCode	—	38.9%
GSO	26.5%	—
ALE-Bench	1,025	—
AlgoTune	1.77	—

Agentic & Tool Use Claude Opus 4.5 leads

Claude Opus 4.5: 47.3 (#12), DeepSeek-V3.2-Exp: 32.7 (#59)

Agentic & Tool Use benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
Terminal-Bench	63.1%	39.6%
Berkeley Function Calling Leaderboard	77.5%	56.7%
Vending-Bench 2	4,967	1,034
APEX-Agents	—	21.3%
GDPval	45.5%	—
Remote Labor Index	3.8%	—
TheAgentCompany	—	42.9%
τ²-bench Airline	84%	—
τ²-bench Banking	24.7%	—
τ²-bench Retail	79.6%	—
τ²-bench Telecom	92.3%	—
Cybench	82%	—
DeepResearch Bench	54.8%	—
OSWorld	66.3%	—
BALROG	43.5%	—
LMArena Search	1180	—
METR Time Horizons	75%	—

Reasoning Claude Opus 4.5 leads

Claude Opus 4.5: 42.6 (#51), DeepSeek-V3.2-Exp: 22.1 (#208)

Reasoning benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
ARC-AGI-2	37.6%	4%
Kagi LLM Benchmark	80.2%	52.2%
NYT Connections (extended)	52.5%	36.7%
ARC-AGI-1	80%	57%
Chess Puzzles	12%	14%
LMArena Hard Prompts	1476	1434
DTBench	89.9%	87.7%
LMCA	44.5%	29.1%
Epoch Capabilities Index	150.09	146.27
SimpleBench	62%	—
CritPt	—	2.9%
EnigmaEval	11.9%	—
Thematic Generalization	—	65%
EBR-Bench	14.3%	—
Mystery Game Puzzles	22%	—
ForecastBench	60.7	—

Math DeepSeek-V3.2-Exp leads

Claude Opus 4.5: 38.6 (#132), DeepSeek-V3.2-Exp: 41.7 (#87)

Math benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
OTIS Mock AIME 2024-2025	86.1%	87.8%
ProofBench	36%	8%
LMArena Math	1463	1435
FrontierMath (Feb 2025 set)	20.7%	22.1%
FrontierMath Tier 4 (v1)	4.2%	2.1%
FrontierMath (Tiers 1-3)	34.4%	—
FrontierMath Tier 4	4.9%	—
MathArena Final-Answer Competitions	—	57.7%

Knowledge Claude Opus 4.5 leads

Claude Opus 4.5: 56.5 (#44), DeepSeek-V3.2-Exp: 51.7 (#66)

Knowledge benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
GPQA Diamond	86%	83.4%
Vectara Hallucination Rate	10.9%	5.3%
LMArena Expert	1487	1436
Humanity's Last Exam	25.2%	—
SimpleQA Verified	45.7%	—

Multimodal Not comparable

Claude Opus 4.5: 31.4 (#107), DeepSeek-V3.2-Exp: —

Multimodal benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
GeoBench	75%	—
VPCT	40%	—
Furniture Assembly	28.3%	—
LMArena Document	1462	—

Multilingual Claude Opus 4.5 leads

Claude Opus 4.5: 54.3 (#47), DeepSeek-V3.2-Exp: 52.2 (#90)

Multilingual benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
LMArena Non-English	1438	1409
LMArena Chinese	1470	1461
LMArena French	1471	1433
LMArena German	1449	1440
LMArena Japanese	1416	1374
LMArena Korean	1424	1371
LMArena Russian	1447	1424
LMArena Spanish	1458	1440

Instruction Following Claude Opus 4.5 leads

Claude Opus 4.5: 77.5 (#19), DeepSeek-V3.2-Exp: 74.5 (#93)

Instruction Following benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
LMArena Instruction Following	1478	1413

Long Context DeepSeek-V3.2-Exp leads

Claude Opus 4.5: 46.5 (#22), DeepSeek-V3.2-Exp: 47.6 (#16)

Long Context benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
CL-bench	21.1%	13.2%
LMArena Longer Query	1480	1428
Fiction.LiveBench	—	83.3%
CL-bench Life	—	9.5%

Writing & Preference Claude Opus 4.5 leads

Claude Opus 4.5: 68.1 (#28), DeepSeek-V3.2-Exp: 62.4 (#77)

Writing & Preference benchmarks
Benchmark	Claude Opus 4.5	DeepSeek-V3.2-Exp
LMArena Text	1451	1425
LMArena Creative Writing	1445	1403
EQ-Bench Creative Writing	1687	1515
LMArena Multi-Turn	1466	1427

Frequently asked questions

Is Claude Opus 4.5 better than DeepSeek-V3.2-Exp?

Which is cheaper, Claude Opus 4.5 or DeepSeek-V3.2-Exp?

DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Claude Opus 4.5 lists at $5 and $25.

Is Claude Opus 4.5 or DeepSeek-V3.2-Exp better for coding?

Claude Opus 4.5 scores higher on coding benchmarks: 54.8 versus 46.5 in the Noometry coding category.

Which has the bigger context window?

Claude Opus 4.5 does, with 200K tokens against 164K.

How many benchmarks do Claude Opus 4.5 and DeepSeek-V3.2-Exp share?

40 benchmarks have published results for both models. Claude Opus 4.5 has 69 scored results on Noometry and DeepSeek-V3.2-Exp has 49.