Noometry

Model comparison

Claude Opus 4.7 vs DeepSeek-V3.2-Exp

Claude Opus 4.7 is the stronger model overall, scoring 58.3 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 4.7's lead doesn't matter for your workload.

Last verified . 41 shared benchmarks.

Claude Opus 4.7 Anthropic

58.3

Rank #19 Confirmed

DeepSeek-V3.2-Exp DeepSeek

44.3

Rank #78 Confirmed

Summary

  • They share 41 benchmarks with published results for both. Claude Opus 4.7 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 4.7 leads 53.8 to 22.1.
  • The biggest single-benchmark swing is ARC-AGI-2: 75.8% for Claude Opus 4.7 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 4.7.
  • Claude Opus 4.7 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 4.7DeepSeek-V3.2-Exp
ProviderAnthropicDeepSeek
Noometry Index58.344.3
Released2026-04-142025-09-29
WeightsProprietaryOpen
Context window1M164K
Max output128K66K
Input $ / M tokens$5$0.26
Output $ / M tokens$25$0.38
Results tracked6649

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

Category by category

Coding Claude Opus 4.7 leads

Claude Opus 4.7: 59.6 (#13), DeepSeek-V3.2-Exp: 46.5 (#65)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena WebDev15581362
SciCode54.5%38.9%
WeirdML76.4%39.5%
LMArena Coding15181454
SWE-bench Verified83.5%—
FrontierCode38.5%—
SWE-bench Verified (bash only)—70%
Aider Polyglot—74.2%
SWE-bench Multilingual—59%
GSO44.1%—
MirrorCode31.1%—
ALE-Bench1,323—

Agentic & Tool Use Claude Opus 4.7 leads

Claude Opus 4.7: 47.9 (#10), DeepSeek-V3.2-Exp: 32.7 (#59)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
Terminal-Bench80.2%39.6%
APEX-Agents49.2%21.3%
Vending-Bench 210,9371,034
Berkeley Function Calling Leaderboard—56.7%
OSWorld 2.018.2%—
TheAgentCompany—42.9%
τ²-bench Banking40.2%—
PostTrainBench28.6%—
ExploitBench26.5%—
GBAEval43.8%—
GDP.pdf21%—
LMArena Search1233—

Reasoning Claude Opus 4.7 leads

Claude Opus 4.7: 53.8 (#29), DeepSeek-V3.2-Exp: 22.1 (#208)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
ARC-AGI-275.8%4%
Kagi LLM Benchmark80.7%52.2%
NYT Connections (extended)39%36.7%
ARC-AGI-193.5%57%
CritPt12%2.9%
Chess Puzzles30%14%
Thematic Generalization72.8%65%
LMArena Hard Prompts15061434
DTBench94.7%87.7%
LMCA52.2%29.1%
Epoch Capabilities Index156.25146.27
SimpleBench61.7%—
EBR-Bench19%—
Mystery Game Puzzles28%—
ForecastBench60.3—

Math Claude Opus 4.7 leads

Claude Opus 4.7: 66.7 (#26), DeepSeek-V3.2-Exp: 41.7 (#87)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
MathArena Final-Answer Competitions73.6%57.7%
OTIS Mock AIME 2024-202597.8%87.8%
ProofBench54%8%
LMArena Math14991435
FrontierMath (Feb 2025 set)43.8%22.1%
FrontierMath Tier 4 (v1)22.9%2.1%
FrontierMath (Tiers 1-3)70.2%—
FrontierMath Tier 431.7%—

Knowledge Claude Opus 4.7 leads

Claude Opus 4.7: 62.6 (#23), DeepSeek-V3.2-Exp: 51.7 (#66)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
GPQA Diamond90.2%83.4%
Vectara Hallucination Rate12%5.3%
LMArena Expert15211436
Humanity's Last Exam36.2%—
SimpleQA Verified51.7%—

Multimodal Not comparable

Claude Opus 4.7: 41.2 (#38), DeepSeek-V3.2-Exp: —

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena Vision1316—
Blueprint-Bench 224.5%—
Furniture Assembly33.3%—
LMArena Document1495—

Multilingual Claude Opus 4.7 leads

Claude Opus 4.7: 57.3 (#10), DeepSeek-V3.2-Exp: 52.2 (#90)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena Non-English14801409
LMArena Chinese15311461
LMArena French15031433
LMArena German14951440
LMArena Japanese14721374
LMArena Korean14641371
LMArena Russian14941424
LMArena Spanish14951440

Instruction Following Claude Opus 4.7 leads

Claude Opus 4.7: 78.4 (#10), DeepSeek-V3.2-Exp: 74.5 (#93)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena Instruction Following14981413

Long Context DeepSeek-V3.2-Exp leads

Claude Opus 4.7: 46.2 (#25), DeepSeek-V3.2-Exp: 47.6 (#16)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena Longer Query15051428
Fiction.LiveBench—83.3%
CL-bench—13.2%
CL-bench Life—9.5%

Writing & Preference Claude Opus 4.7 leads

Claude Opus 4.7: 75.1 (#8), DeepSeek-V3.2-Exp: 62.4 (#77)

BenchmarkClaude Opus 4.7DeepSeek-V3.2-Exp
LMArena Text14901425
LMArena Creative Writing14861403
EQ-Bench Creative Writing19141515
LMArena Multi-Turn15051427
EQ-Bench 41311—

Frequently asked questions

Is Claude Opus 4.7 better than DeepSeek-V3.2-Exp?

Claude Opus 4.7 is the stronger model overall, scoring 58.3 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 4.7's lead doesn't matter for your workload.

Which is cheaper, Claude Opus 4.7 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 4.7 lists at $5 and $25.

Is Claude Opus 4.7 or DeepSeek-V3.2-Exp better for coding?

Claude Opus 4.7 scores higher on coding benchmarks: 59.6 versus 46.5 in the Noometry coding category.

Which has the bigger context window?

Claude Opus 4.7 does, with 1M tokens against 164K.

How many benchmarks do Claude Opus 4.7 and DeepSeek-V3.2-Exp share?

41 benchmarks have published results for both models. Claude Opus 4.7 has 66 scored results on Noometry and DeepSeek-V3.2-Exp has 49.

Related comparisons

Go deeper