Model comparison

Claude Opus 5 vs DeepSeek-V3.2-Exp

Claude Opus 5 is the stronger model overall, scoring 67.8 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 5's lead doesn't matter for your workload.

Last verified October 10, 2026. 34 shared benchmarks.

Claude Opus 5 Anthropic

67.8

Rank #4 Confirmed

DeepSeek-V3.2-Exp

DeepSeek

44.3

Rank #78 Confirmed

Summary

They share 34 benchmarks with published results for both. Claude Opus 5 scores higher in 8 categories and DeepSeek-V3.2-Exp in 1 category; 9 gaps are clear of the uncertainty.
The widest gap is in reasoning, where Claude Opus 5 leads 77.2 to 22.1.
The biggest single-benchmark swing is ProofBench: 99% for Claude Opus 5 and 8% 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 5.
Claude Opus 5 accepts more context: 1M tokens versus 164K.
DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.

Side by side

Claude Opus 5 and DeepSeek-V3.2-Exp specifications
	Claude Opus 5	DeepSeek-V3.2-Exp
Provider	Anthropic	DeepSeek
Noometry Index	67.8	44.3
Released	2026-07-24	2025-09-29
Weights	Proprietary	Open
Context window	1M	164K
Max output	128K	66K
Input $ / M tokens	$5	$0.26
Output $ / M tokens	$25	$0.38
Results tracked	57	49

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

Category by category

Coding Claude Opus 5 leads

Claude Opus 5: 67.5 (#5), DeepSeek-V3.2-Exp: 46.5 (#65)

Coding benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena WebDev	1691	1362
SciCode	56.4%	38.9%
WeirdML	91.8%	39.5%
LMArena Coding	1534	1454
DeepSWE	73.6%	—
FrontierCode	53.4%	—
SWE-bench Verified (bash only)	—	70%
Aider Polyglot	—	74.2%
CursorBench	46.6%	—
SWE-bench Multilingual	—	59%
FrontierSWE	52%	—
ALE-Bench	2,165	—

Agentic & Tool Use Claude Opus 5 leads

Claude Opus 5: 55.6 (#1), DeepSeek-V3.2-Exp: 32.7 (#59)

Agentic & Tool Use benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
APEX-Agents	65.8%	21.3%
Vending-Bench 2	11,182	1,034
Terminal-Bench	—	39.6%
Berkeley Function Calling Leaderboard	—	56.7%
OSWorld 2.0	31.4%	—
TheAgentCompany	—	42.9%
τ²-bench Banking	48.7%	—
PostTrainBench	35%	—
BALROG	63.4%	—
GBAEval	79.6%	—
GDP.pdf	24%	—

Reasoning Claude Opus 5 leads

Claude Opus 5: 77.2 (#4), DeepSeek-V3.2-Exp: 22.1 (#208)

Reasoning benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
ARC-AGI-2	90.4%	4%
NYT Connections (extended)	94.3%	36.7%
ARC-AGI-1	97.5%	57%
CritPt	29.1%	2.9%
Chess Puzzles	42%	14%
LMArena Hard Prompts	1526	1434
DTBench	97.9%	87.7%
LMCA	64.5%	29.1%
Epoch Capabilities Index	162.78	146.27
SimpleBench	80.6%	—
Kagi LLM Benchmark	—	52.2%
Thematic Generalization	—	65%
EBR-Bench	45.7%	—
Mystery Game Puzzles	59%	—
Bench to the Future 3	0.12	—

Math Claude Opus 5 leads

Claude Opus 5: 86.2 (#8), DeepSeek-V3.2-Exp: 41.7 (#87)

Math benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
OTIS Mock AIME 2024-2025	98.9%	87.8%
ProofBench	99%	8%
LMArena Math	1531	1435
FrontierMath (Tiers 1-3)	85.6%	—
FrontierMath Tier 4	73.2%	—
MathArena Final-Answer Competitions	—	57.7%
FrontierMath (Feb 2025 set)	—	22.1%
FrontierMath Tier 4 (v1)	—	2.1%

Knowledge Claude Opus 5 leads

Claude Opus 5: 66.8 (#9), DeepSeek-V3.2-Exp: 51.7 (#66)

Knowledge benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
GPQA Diamond	93.9%	83.4%
LMArena Expert	1557	1436
SimpleQA Verified	59.9%	—
Vectara Hallucination Rate	—	5.3%

Multimodal Not comparable

Claude Opus 5: 50.8 (#8), DeepSeek-V3.2-Exp: —

Multimodal benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena Vision	1319	—
Blueprint-Bench 2	30.4%	—
Furniture Assembly	60.8%	—
LMArena Document	1516	—

Multilingual Claude Opus 5 leads

Claude Opus 5: 58.8 (#4), DeepSeek-V3.2-Exp: 52.2 (#90)

Multilingual benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena Non-English	1501	1409
LMArena Chinese	1574	1461
LMArena French	1519	1433
LMArena German	1524	1440
LMArena Japanese	1516	1374
LMArena Korean	1521	1371
LMArena Russian	1507	1424
LMArena Spanish	1519	1440

Instruction Following Claude Opus 5 leads

Claude Opus 5: 79.2 (#7), DeepSeek-V3.2-Exp: 74.5 (#93)

Instruction Following benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena Instruction Following	1517	1413

Long Context DeepSeek-V3.2-Exp leads

Claude Opus 5: 46.5 (#21), DeepSeek-V3.2-Exp: 47.6 (#16)

Long Context benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena Longer Query	1515	1428
Fiction.LiveBench	—	83.3%
CL-bench	—	13.2%
CL-bench Life	—	9.5%

Writing & Preference Claude Opus 5 leads

Claude Opus 5: 79.2 (#1), DeepSeek-V3.2-Exp: 62.4 (#77)

Writing & Preference benchmarks
Benchmark	Claude Opus 5	DeepSeek-V3.2-Exp
LMArena Text	1507	1425
LMArena Creative Writing	1491	1403
EQ-Bench Creative Writing	2133	1515
LMArena Multi-Turn	1499	1427
EQ-Bench 4	1385	—

Frequently asked questions

Is Claude Opus 5 better than DeepSeek-V3.2-Exp?

Which is cheaper, Claude Opus 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 5 lists at $5 and $25.

Is Claude Opus 5 or DeepSeek-V3.2-Exp better for coding?

Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 46.5 in the Noometry coding category.

Which has the bigger context window?

Claude Opus 5 does, with 1M tokens against 164K.

How many benchmarks do Claude Opus 5 and DeepSeek-V3.2-Exp share?

34 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and DeepSeek-V3.2-Exp has 49.