Noometry

Model comparison

DeepSeek-V3.2-Exp vs Step 3.7 Flash

DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 37.3 on the Noometry Index.

Last verified . 4 shared benchmarks.

DeepSeek-V3.2-Exp DeepSeek

44.3

Rank #78 Confirmed

Step 3.7 Flash StepFun

37.3

Rank #207 Reported

Summary

  • They share 4 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 2 categories and Step 3.7 Flash in 1 category; 2 gaps are clear of the uncertainty.
  • The widest gap is in coding, where DeepSeek-V3.2-Exp leads 46.5 to 40.0.
  • The biggest single-benchmark swing is MathArena Final-Answer Competitions: 57.7% for DeepSeek-V3.2-Exp and 68.5% for Step 3.7 Flash.
  • DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.18 / $1.11 for Step 3.7 Flash.
  • Step 3.7 Flash accepts more context: 256K tokens versus 164K.

Side by side

DeepSeek-V3.2-ExpStep 3.7 Flash
ProviderDeepSeekStepFun
Noometry Index44.337.3
Released2025-09-292026-05-29
WeightsOpenOpen
Context window164K256K
Max output66K256K
Input $ / M tokens$0.26$0.18
Output $ / M tokens$0.38$1.11
Results tracked495

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

Category by category

Coding DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 46.5 (#65), Step 3.7 Flash: 40.0 (#150)

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
SciCode38.9%40%
SWE-bench Verified (bash only)70%—
Aider Polyglot74.2%—
LMArena WebDev1362—
SWE-bench Multilingual59%—
WeirdML39.5%—
LMArena Coding1454—
ALE-Bench—694.12

Agentic & Tool Use Not comparable

DeepSeek-V3.2-Exp: 32.7 (#59), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
Terminal-Bench39.6%—
APEX-Agents21.3%—
Berkeley Function Calling Leaderboard56.7%—
TheAgentCompany42.9%—
Vending-Bench 21,034—

Reasoning Too close to call

DeepSeek-V3.2-Exp: 22.1 (#208), Step 3.7 Flash: 21.6 (#219)

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
NYT Connections (extended)36.7%39.7%
CritPt2.9%2.3%
ARC-AGI-24%—
Kagi LLM Benchmark52.2%—
ARC-AGI-157%—
Chess Puzzles14%—
Thematic Generalization65%—
LMArena Hard Prompts1434—
DTBench87.7%—
LMCA29.1%—
Epoch Capabilities Index146.27—

Math Step 3.7 Flash leads

DeepSeek-V3.2-Exp: 41.7 (#87), Step 3.7 Flash: 42.9 (#82)

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
MathArena Final-Answer Competitions57.7%68.5%
OTIS Mock AIME 2024-202587.8%—
ProofBench8%—
LMArena Math1435—
FrontierMath (Feb 2025 set)22.1%—
FrontierMath Tier 4 (v1)2.1%—

Knowledge Not comparable

DeepSeek-V3.2-Exp: 51.7 (#66), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
GPQA Diamond83.4%—
Vectara Hallucination Rate5.3%—
LMArena Expert1436—

Multilingual Not comparable

DeepSeek-V3.2-Exp: 52.2 (#90), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
LMArena Non-English1409—
LMArena Chinese1461—
LMArena French1433—
LMArena German1440—
LMArena Japanese1374—
LMArena Korean1371—
LMArena Russian1424—
LMArena Spanish1440—

Instruction Following Not comparable

DeepSeek-V3.2-Exp: 74.5 (#93), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
LMArena Instruction Following1413—

Long Context Not comparable

DeepSeek-V3.2-Exp: 47.6 (#16), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
Fiction.LiveBench83.3%—
CL-bench13.2%—
CL-bench Life9.5%—
LMArena Longer Query1428—

Writing & Preference Not comparable

DeepSeek-V3.2-Exp: 62.4 (#77), Step 3.7 Flash: —

BenchmarkDeepSeek-V3.2-ExpStep 3.7 Flash
LMArena Text1425—
LMArena Creative Writing1403—
EQ-Bench Creative Writing1515—
LMArena Multi-Turn1427—

Frequently asked questions

Is DeepSeek-V3.2-Exp better than Step 3.7 Flash?

DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 37.3 on the Noometry Index.

Which is cheaper, DeepSeek-V3.2-Exp or Step 3.7 Flash?

DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Step 3.7 Flash lists at $0.18 and $1.11.

Is DeepSeek-V3.2-Exp or Step 3.7 Flash better for coding?

DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 40.0 in the Noometry coding category.

Which has the bigger context window?

Step 3.7 Flash does, with 256K tokens against 164K.

How many benchmarks do DeepSeek-V3.2-Exp and Step 3.7 Flash share?

4 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and Step 3.7 Flash has 5.

Related comparisons

Go deeper