Noometry

Model comparison

DeepSeek V4 Flash vs Step 3.7 Flash

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 37.3 on the Noometry Index.

Last verified . 5 shared benchmarks.

DeepSeek V4 Flash DeepSeek

53.6

Rank #35 Confirmed

Step 3.7 Flash StepFun

37.3

Rank #207 Reported

Summary

  • They share 5 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 3 categories and Step 3.7 Flash in 0 categories; 3 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 21.6.
  • The biggest single-benchmark swing is NYT Connections (extended): 89.6% for DeepSeek V4 Flash and 39.7% for Step 3.7 Flash.
  • DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.18 / $1.11 for Step 3.7 Flash.
  • DeepSeek V4 Flash accepts more context: 1M tokens versus 256K.

Side by side

DeepSeek V4 FlashStep 3.7 Flash
ProviderDeepSeekStepFun
Noometry Index53.637.3
Released2026-04-242026-05-29
WeightsOpenOpen
Context window1M256K
Max output393K256K
Input $ / M tokens$0.15$0.18
Output $ / M tokens$0.60$1.11
Results tracked415

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

Category by category

Coding DeepSeek V4 Flash leads

DeepSeek V4 Flash: 47.9 (#59), Step 3.7 Flash: 40.0 (#150)

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
SciCode49.9%40%
ALE-Bench1,306694.12
FrontierCode18.8%—
LMArena WebDev1582—
WeirdML63%—
LMArena Coding1457—

Reasoning DeepSeek V4 Flash leads

DeepSeek V4 Flash: 53.7 (#30), Step 3.7 Flash: 21.6 (#219)

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
NYT Connections (extended)89.6%39.7%
CritPt16.6%2.3%
ARC-AGI-261.4%—
SimpleBench61.1%—
Kagi LLM Benchmark52.2%—
ARC-AGI-189%—
Chess Puzzles33%—
LMArena Hard Prompts1444—
Mystery Game Puzzles34%—
DTBench90.9%—
LMCA41.7%—
Epoch Capabilities Index154.49—

Math DeepSeek V4 Flash leads

DeepSeek V4 Flash: 60.3 (#37), Step 3.7 Flash: 42.9 (#82)

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
MathArena Final-Answer Competitions76.5%68.5%
FrontierMath (Tiers 1-3)57.5%—
FrontierMath Tier 424.4%—
OTIS Mock AIME 2024-202594.4%—
ProofBench56%—
LMArena Math1427—

Knowledge Not comparable

DeepSeek V4 Flash: 55.4 (#48), Step 3.7 Flash: —

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
GPQA Diamond91%—
SimpleQA Verified33.6%—
LMArena Expert1441—

Multilingual Not comparable

DeepSeek V4 Flash: 53.0 (#72), Step 3.7 Flash: —

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
LMArena Non-English1420—
LMArena Chinese1468—
LMArena French1439—
LMArena German1418—
LMArena Japanese1406—
LMArena Korean1384—
LMArena Russian1428—
LMArena Spanish1436—

Instruction Following Not comparable

DeepSeek V4 Flash: 74.9 (#81), Step 3.7 Flash: —

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
LMArena Instruction Following1421—

Long Context Not comparable

DeepSeek V4 Flash: 43.8 (#85), Step 3.7 Flash: —

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
LMArena Longer Query1434—

Writing & Preference Not comparable

DeepSeek V4 Flash: 63.8 (#61), Step 3.7 Flash: —

BenchmarkDeepSeek V4 FlashStep 3.7 Flash
LMArena Text1432—
LMArena Creative Writing1403—
EQ-Bench Creative Writing1559—
LMArena Multi-Turn1449—

Frequently asked questions

Is DeepSeek V4 Flash better than Step 3.7 Flash?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 37.3 on the Noometry Index.

Which is cheaper, DeepSeek V4 Flash or Step 3.7 Flash?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Step 3.7 Flash lists at $0.18 and $1.11.

Is DeepSeek V4 Flash or Step 3.7 Flash better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 40.0 in the Noometry coding category.

Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 256K.

How many benchmarks do DeepSeek V4 Flash and Step 3.7 Flash share?

5 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Step 3.7 Flash has 5.

Related comparisons

Go deeper