Enter opens the highlighted result. Esc closes.

DeepSeek V4.1 Flash vs Qwen3.5 397B A17B

DeepSeek V4.1 Flash leads Qwen3.5 397B A17B on the Artificial Analysis Intelligence Index, 24.7 to 21.4, and DeepSeek V4.1 Flash costs 2.8x less per output token ($0.84 vs $2.34 per million). Updated October 22, 2026; 59 benchmarks have a result for both.

Replace DeepSeek V4.1 Flash Replace Qwen3.5 397B A17B
DeepSeek V4.1 FlashDeepSeek
Coding
62.5
Overall
62.4
Output $/M
$0.84
58
Qwen3.5 397B A17BQwen
Coding
No data
Overall
57.1
Output $/M
$2.34
1
vs boards won of 59 shared

DeepSeek V4.1 Flash or Qwen3.5 397B A17B: which to pick

Only DeepSeek V4.1 Flash has a coding score, 62.5 on the Vibe Coding Index; Qwen3.5 397B A17B has none, so the two cannot be ranked on coding.

Pick Qwen3.5 397B A17B for work measured by LMArena Vision: humor, where it leads.

Model ADeepSeek V4.1 FlashDeepSeek62.5/1005 independent benchmarksNo cost per solved taskModel BQwen3.5 397B A17BQwenNot benchmarked1 independent benchmarkNo cost per solved task
Shared results
59
DeepSeek V4.1 Flash only
42
Qwen3.5 397B A17B only
6
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
DeepSeek V4.1 Flash 62.5 from 5 benchmarks, Qwen3.5 397B A17B not benchmarked.
Widest category
Coding, 19.0 points apart on average over 7 shared benchmarks.
Widest single gap
FrontierMath, tiers 1 to 3: DeepSeek V4.1 Flash 67.4% against Qwen3.5 397B A17B 31.2%, 36.1 points apart on a 0 to 100 scale.
Output price
DeepSeek V4.1 Flash $0.84 against Qwen3.5 397B A17B $2.34 per million output tokens, 2.8x apart.
Context window
DeepSeek V4.1 Flash takes 1.05M tokens, Qwen3.5 397B A17B 262K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryDeepSeek V4.1 FlashBarsQwen3.5 397B A17BTies
Coding70
Intelligence20
Agentic work10
Reasoning80
Math40
Knowledge40
Web design70
General251
All shared boards581
CodingLike-for-likeSix of these boards feed the Vibe Coding Index70
Coding benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
LiveBench coding90.2%Not measuredOnly DeepSeek V4.1 Flash measured
LMArena WebDev16181400DeepSeek V4.1 Flash
Terminal-Bench 4 (Vals AI)19.7%Not measuredOnly DeepSeek V4.1 Flash measured
Vibe Code Bench (Vals AI)84.7%Not measuredOnly DeepSeek V4.1 Flash measured
Code migration (Vals AI)45.6%Not measuredOnly DeepSeek V4.1 Flash measured
ProgramBench (Vals AI)0.5%Not measuredOnly DeepSeek V4.1 Flash measured
IOI (Vals AI)40.3%Not measuredOnly DeepSeek V4.1 Flash measured
Terminal-Bench 2.1 (Vals AI)74.5%Not measuredOnly DeepSeek V4.1 Flash measured
SWE-bench Multimodal (Vals AI)46.5%Not measuredOnly DeepSeek V4.1 Flash measured
Vibe Code Bench 1-100 (Vals AI)16.4%Not measuredOnly DeepSeek V4.1 Flash measured
SkillsBench (Vals AI)69.8%Not measuredOnly DeepSeek V4.1 Flash measured
CyberBench (Vals AI)73.7%Not measuredOnly DeepSeek V4.1 Flash measured
SciCode51.9%44.8%DeepSeek V4.1 Flash
Terminal-Bench Hard (Artificial Analysis run)Not measured40.9%Only Qwen3.5 397B A17B measured
FrontierCode43.9%Not measuredOnly DeepSeek V4.1 Flash measured
ALE-Bench1092.3621.9DeepSeek V4.1 Flash
LMArena Text: coding15021464DeepSeek V4.1 Flash
LMArena Text: software and IT14961463DeepSeek V4.1 Flash
LMArena WebDev: React16281388DeepSeek V4.1 Flash
LMArena WebDev: HTML16121410DeepSeek V4.1 Flash
LMArena WebDev: from an image1606Not measuredOnly DeepSeek V4.1 Flash measured
Neither model has a result on the other 18 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
AA Intelligence Index24.721.4DeepSeek V4.1 Flash
Epoch Capabilities Index155.9146.8DeepSeek V4.1 Flash
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
LiveBench research86.8%Not measuredOnly DeepSeek V4.1 Flash measured
LiveBench browser84.0%Not measuredOnly DeepSeek V4.1 Flash measured
LiveBench long agentic40.1%Not measuredOnly DeepSeek V4.1 Flash measured
SREBench (Vals AI)0.8%Not measuredOnly DeepSeek V4.1 Flash measured
Finance Agent (Vals AI)53.5%Not measuredOnly DeepSeek V4.1 Flash measured
Tax Agent Bench (Vals AI)26.6%Not measuredOnly DeepSeek V4.1 Flash measured
Legal Research Bench (Vals AI)41.4%Not measuredOnly DeepSeek V4.1 Flash measured
Excel Modeling Benchmark (Vals AI)57.2%Not measuredOnly DeepSeek V4.1 Flash measured
BioMysteryBench (Vals AI)67.8%Not measuredOnly DeepSeek V4.1 Flash measured
Terminal-Bench-Science (Vals AI)0.0%Not measuredOnly DeepSeek V4.1 Flash measured
Vals Science Index29.6%Not measuredOnly DeepSeek V4.1 Flash measured
τ²-Bench TelecomNot measured95.6%Only Qwen3.5 397B A17B measured
APEX-Agents39.5%24.9%DeepSeek V4.1 Flash
Neither model has a result on the other 28 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index80
Reasoning benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
LiveBench reasoning86.7%Not measuredOnly DeepSeek V4.1 Flash measured
MysteryMechanism (Vals AI)21.2%Not measuredOnly DeepSeek V4.1 Flash measured
GPQA Diamond (Artificial Analysis run)Not measured89.3%Only Qwen3.5 397B A17B measured
Humanity's Last Exam (Artificial Analysis run)39.2%29.0%DeepSeek V4.1 Flash
GPQA Diamond (Epoch AI run)89.8%86.4%DeepSeek V4.1 Flash
ARC-AGI94.5%Not measuredOnly DeepSeek V4.1 Flash measured
ARC-AGI-272.9%Not measuredOnly DeepSeek V4.1 Flash measured
Chess puzzles48.0%Not measuredOnly DeepSeek V4.1 Flash measured
Mystery game puzzles43.0%18.0%DeepSeek V4.1 Flash
LMArena Text: hard prompts14831448DeepSeek V4.1 Flash
LMArena Text: expert15011462DeepSeek V4.1 Flash
LMArena Text: hard prompts in English14881450DeepSeek V4.1 Flash
LMArena Vision: diagrams12971277DeepSeek V4.1 Flash
LMArena Vision: homework13071292DeepSeek V4.1 Flash
Neither model has a result on the other 8 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index40
Math benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
OTIS Mock AIME98.3%88.9%DeepSeek V4.1 Flash
FrontierMath, tiers 1 to 367.4%31.2%DeepSeek V4.1 Flash
FrontierMath, tier 426.8%Not measuredOnly DeepSeek V4.1 Flash measured
ProofBench54.0%Not measuredOnly DeepSeek V4.1 Flash measured
LMArena Text: math14781453DeepSeek V4.1 Flash
LMArena Text: mathematical work14851454DeepSeek V4.1 Flash
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
Knowledge benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
LegalBench (Vals AI)83.3%Not measuredOnly DeepSeek V4.1 Flash measured
MedCode (Vals AI)41.2%Not measuredOnly DeepSeek V4.1 Flash measured
SimpleQA Verified42.9%Not measuredOnly DeepSeek V4.1 Flash measured
LMArena Text: science14691457DeepSeek V4.1 Flash
LMArena Text: medicine and healthcare14841455DeepSeek V4.1 Flash
LMArena Text: legal and government14431441DeepSeek V4.1 Flash
LMArena Text: business and finance14611435DeepSeek V4.1 Flash
Neither model has a result on the other 11 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index70
Web design benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
Design Arena: websites13081190DeepSeek V4.1 Flash
Design Arena: UI components13121160DeepSeek V4.1 Flash
Design Arena: game dev13301140DeepSeek V4.1 Flash
Design Arena: data viz12911175DeepSeek V4.1 Flash
Design Arena: 3D13551160DeepSeek V4.1 Flash
Design Arena: all code categories13201179DeepSeek V4.1 Flash
Design Arena: SVG12451139DeepSeek V4.1 Flash
Design Arena: ASCII art1265Not measuredOnly DeepSeek V4.1 Flash measured
LiveBench design79.2%Not measuredOnly DeepSeek V4.1 Flash measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index251
General benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
LiveBench data analysis79.3%Not measuredOnly DeepSeek V4.1 Flash measured
Public Benefits Bench (Vals AI)64.3%Not measuredOnly DeepSeek V4.1 Flash measured
SAGE (Vals AI)47.9%Not measuredOnly DeepSeek V4.1 Flash measured
MedScribe (Vals AI)85.5%Not measuredOnly DeepSeek V4.1 Flash measured
IFBenchNot measured78.8%Only Qwen3.5 397B A17B measured
Long context reasoning (AA-LCR)84.0%77.3%DeepSeek V4.1 Flash
DTBench89.9%87.5%DeepSeek V4.1 Flash
LMCA47.0%37.9%DeepSeek V4.1 Flash
Surface Evolver Bench46.3%Not measuredOnly DeepSeek V4.1 Flash measured
Furniture assembly34.2%Not measuredOnly DeepSeek V4.1 Flash measured
LMArena Text14641437DeepSeek V4.1 Flash
LMArena Text: instruction following14731424DeepSeek V4.1 Flash
LMArena Text: creative writing14371402DeepSeek V4.1 Flash
LMArena Vision12781262DeepSeek V4.1 Flash
LMArena Text: English14731443DeepSeek V4.1 Flash
LMArena Text: non-English14491430DeepSeek V4.1 Flash
LMArena Text: Chinese15011498DeepSeek V4.1 Flash
LMArena Text: German14701445DeepSeek V4.1 Flash
LMArena Text: Japanese14321429DeepSeek V4.1 Flash
LMArena Text: Korean14361384DeepSeek V4.1 Flash
LMArena Text: French14651462DeepSeek V4.1 Flash
LMArena Text: Spanish14591441DeepSeek V4.1 Flash
LMArena Text: Russian14671430DeepSeek V4.1 Flash
LMArena Text: Polish14571447DeepSeek V4.1 Flash
LMArena Text: multi-turn14611446DeepSeek V4.1 Flash
LMArena Text: longer queries14731441DeepSeek V4.1 Flash
LMArena Text: writing and language14501415DeepSeek V4.1 Flash
LMArena Text: entertainment and media14451399DeepSeek V4.1 Flash
LMArena Vision: English12881268DeepSeek V4.1 Flash
LMArena Vision: Chinese13271314DeepSeek V4.1 Flash
LMArena Vision: reading text12821271DeepSeek V4.1 Flash
LMArena Vision: humor12601262Qwen3.5 397B A17B
LMArena Vision: captioningNot measured1230Only Qwen3.5 397B A17B measured
LMArena Vision: recognizing thingsNot measured1265Only Qwen3.5 397B A17B measured
LMArena Vision: creative writing12751260DeepSeek V4.1 Flash
Neither model has a result on the other 11 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V4.1 Flash and Qwen3.5 397B A17B
BenchmarkDeepSeek V4.1 FlashQwen3.5 397B A17BLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

DeepSeek V4.1 Flash$0.0006Fits in one request
Qwen3.5 397B A17B$0.0016Fits in one request

DeepSeek V4.1 Flash costs less

Repository review

50K input + 3K output tokens

DeepSeek V4.1 Flash$0.014Fits in one request
Qwen3.5 397B A17B$0.027Fits in one request

DeepSeek V4.1 Flash costs less

Agent loop

200K input + 10K output tokens

DeepSeek V4.1 Flash$0.053Fits in one request
Qwen3.5 397B A17B$0.101Fits in one request

DeepSeek V4.1 Flash costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of DeepSeek V4.1 Flash and Qwen3.5 397B A17B
SpecificationDeepSeek V4.1 FlashQwen3.5 397B A17B
MakerDeepSeekQwen
WeightsOpen weightsOpen weights
ReleasedSeptember 10, 2026February 16, 2026
Context window1.05M262K
Input $/Mtok$0.2225$0.39
Output $/Mtok$0.84$2.34
Coding index62.5Not benchmarked
Independent measurements behind it51
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V4.1 Flash better than Qwen3.5 397B A17B for coding?

Only DeepSeek V4.1 Flash has a coding score, 62.5 on the Vibe Coding Index; Qwen3.5 397B A17B has none, so the two cannot be ranked on coding. Pick Qwen3.5 397B A17B for work measured by LMArena Vision: humor, where it leads.

Which is cheaper, DeepSeek V4.1 Flash or Qwen3.5 397B A17B?

DeepSeek V4.1 Flash costs $0.2225 per million input tokens and $0.84 per million output tokens; Qwen3.5 397B A17B costs $0.39 and $2.34. Cost per solved task is known for neither of them.

Which has the bigger context window, DeepSeek V4.1 Flash or Qwen3.5 397B A17B?

DeepSeek V4.1 Flash, with 1.05M tokens against 262K.

How many benchmarks is the DeepSeek V4.1 Flash vs Qwen3.5 397B A17B comparison based on?

59 benchmarks have a result for both models as of October 22, 2026. DeepSeek V4.1 Flash's coding index rests on 5 independent benchmarks and Qwen3.5 397B A17B's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next