Noometry

Model comparison

Claude Opus 4.6 vs DeepSeek-V3.2-Exp

Claude Opus 4.6 is the stronger model overall, scoring 58.2 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 4.6's lead doesn't matter for your workload.

Last verified . 43 shared benchmarks.

Claude Opus 4.6 Anthropic

58.2

Rank #20 Confirmed

DeepSeek-V3.2-Exp DeepSeek

44.3

Rank #78 Confirmed

Summary

  • They share 43 benchmarks with published results for both. Claude Opus 4.6 scores higher in 9 categories and DeepSeek-V3.2-Exp in 0 categories; 8 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where Claude Opus 4.6 leads 57.8 to 22.1.
  • The biggest single-benchmark swing is ARC-AGI-2: 69.2% for Claude Opus 4.6 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 4.6.
  • Claude Opus 4.6 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 4.6DeepSeek-V3.2-Exp
ProviderAnthropicDeepSeek
Noometry Index58.244.3
Released2026-02-042025-09-29
WeightsProprietaryOpen
Context window1M164K
Max output128K66K
Input $ / M tokens$5$0.26
Output $ / M tokens$25$0.38
Results tracked6849

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

Category by category

Coding Claude Opus 4.6 leads

Claude Opus 4.6: 57.2 (#20), DeepSeek-V3.2-Exp: 46.5 (#65)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
SWE-bench Verified (bash only)75.6%70%
LMArena WebDev15471362
SWE-bench Multilingual72%59%
WeirdML78%39.5%
LMArena Coding15361454
SWE-bench Verified78.7%—
FrontierCode26.6%—
Aider Polyglot—74.2%
SciCode—38.9%
GSO41.2%—
ALE-Bench996.5—
AlgoTune1.47—

Agentic & Tool Use Claude Opus 4.6 leads

Claude Opus 4.6: 51.1 (#4), DeepSeek-V3.2-Exp: 32.7 (#59)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
Terminal-Bench79.8%39.6%
APEX-Agents46.3%21.3%
Vending-Bench 28,0181,034
Berkeley Function Calling Leaderboard—56.7%
Remote Labor Index4.2%—
TheAgentCompany—42.9%
τ²-bench Banking27.3%—
Cybench93%—
DeepResearch Bench55.3%—
GBAEval44.1%—
LMArena Search1253—
METR Time Horizons78.9%—

Reasoning Claude Opus 4.6 leads

Claude Opus 4.6: 57.8 (#23), DeepSeek-V3.2-Exp: 22.1 (#208)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
ARC-AGI-269.2%4%
Kagi LLM Benchmark83.6%52.2%
NYT Connections (extended)92.1%36.7%
ARC-AGI-194%57%
Chess Puzzles17%14%
Thematic Generalization80.6%65%
LMArena Hard Prompts15271434
DTBench91.2%87.7%
LMCA55.8%29.1%
Epoch Capabilities Index155.24146.27
SimpleBench67.6%—
CritPt—2.9%
EnigmaEval7.6%—
EBR-Bench12.7%—
Mystery Game Puzzles25%—
ForecastBench60—

Math Claude Opus 4.6 leads

Claude Opus 4.6: 63.0 (#31), DeepSeek-V3.2-Exp: 41.7 (#87)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
MathArena Final-Answer Competitions78.5%57.7%
OTIS Mock AIME 2024-202594.4%87.8%
ProofBench50%8%
LMArena Math15191435
FrontierMath (Feb 2025 set)40.7%22.1%
FrontierMath Tier 4 (v1)22.9%2.1%
FrontierMath (Tiers 1-3)66%—
FrontierMath Tier 426.8%—

Knowledge Claude Opus 4.6 leads

Claude Opus 4.6: 61.9 (#26), DeepSeek-V3.2-Exp: 51.7 (#66)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
GPQA Diamond90.5%83.4%
Vectara Hallucination Rate12.2%5.3%
LMArena Expert15461436
Humanity's Last Exam34.4%—
SimpleQA Verified47%—

Multimodal Not comparable

Claude Opus 4.6: 37.3 (#74), DeepSeek-V3.2-Exp: —

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
LMArena Vision1316—
Furniture Assembly28.3%—
LMArena Document1507—

Multilingual Claude Opus 4.6 leads

Claude Opus 4.6: 57.9 (#6), DeepSeek-V3.2-Exp: 52.2 (#90)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
LMArena Non-English14891409
LMArena Chinese15511461
LMArena French15131433
LMArena German15021440
LMArena Japanese14841374
LMArena Korean14641371
LMArena Russian14971424
LMArena Spanish15101440

Instruction Following Claude Opus 4.6 leads

Claude Opus 4.6: 79.5 (#4), DeepSeek-V3.2-Exp: 74.5 (#93)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
LMArena Instruction Following15231413

Long Context Too close to call

Claude Opus 4.6: 48.1 (#13), DeepSeek-V3.2-Exp: 47.6 (#16)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
CL-bench20.7%13.2%
CL-bench Life17%9.5%
LMArena Longer Query15201428
Fiction.LiveBench—83.3%

Writing & Preference Claude Opus 4.6 leads

Claude Opus 4.6: 73.5 (#10), DeepSeek-V3.2-Exp: 62.4 (#77)

BenchmarkClaude Opus 4.6DeepSeek-V3.2-Exp
LMArena Text15031425
LMArena Creative Writing15051403
EQ-Bench Creative Writing18091515
LMArena Multi-Turn15131427
EQ-Bench 41223—

Frequently asked questions

Is Claude Opus 4.6 better than DeepSeek-V3.2-Exp?

Claude Opus 4.6 is the stronger model overall, scoring 58.2 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 4.6's lead doesn't matter for your workload.

Which is cheaper, Claude Opus 4.6 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 4.6 lists at $5 and $25.

Is Claude Opus 4.6 or DeepSeek-V3.2-Exp better for coding?

Claude Opus 4.6 scores higher on coding benchmarks: 57.2 versus 46.5 in the Noometry coding category.

Which has the bigger context window?

Claude Opus 4.6 does, with 1M tokens against 164K.

How many benchmarks do Claude Opus 4.6 and DeepSeek-V3.2-Exp share?

43 benchmarks have published results for both models. Claude Opus 4.6 has 68 scored results on Noometry and DeepSeek-V3.2-Exp has 49.

Related comparisons

Go deeper