Enter opens the highlighted result. Esc closes.

DeepSeek V4 Flash 0731 vs GLM 5.2

GLM 5.2 leads DeepSeek V4 Flash 0731 on coding, 61.1 to 51.8 on the Vibe Coding Index, and DeepSeek V4 Flash 0731 costs 15.7x less per output token ($0.28 vs $4.40 per million). Updated October 21, 2026; 49 benchmarks have a result for both.

Replace DeepSeek V4 Flash 0731 Replace GLM 5.2
DeepSeek V4 Flash 0731DeepSeek
Coding
51.8
Overall
55.2
Output $/M
$0.28
21
GLM 5.2Z.ai
Coding
61.1
Overall
62.5
Output $/M
$4.40
28
vs boards won of 49 shared

DeepSeek V4 Flash 0731 or GLM 5.2: which to pick

Pick GLM 5.2 for coding: it scores 61.1 on the Vibe Coding Index against 51.8 for DeepSeek V4 Flash 0731.

Pick DeepSeek V4 Flash 0731 for output tokens 15.7x cheaper ($0.28 vs $4.40 per million) and a lower cost per solved task ($0.01 against $0.86).

Model ADeepSeek V4 Flash 0731DeepSeek51.8/1004 independent benchmarks$0.01 per solved taskModel BGLM 5.2Z.ai61.1/1004 independent benchmarks$0.86 per solved task
Shared results
49
DeepSeek V4 Flash 0731 only
6
GLM 5.2 only
49
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 Flash 0731 51.8 from 4 benchmarks, GLM 5.2 61.1 from 4 benchmarks.
Widest category
Reasoning, 9.6 points apart on average over 9 shared benchmarks.
Widest single gap
ARC-AGI-2: DeepSeek V4 Flash 0731 61.4% against GLM 5.2 22.8%, 38.6 points apart on a 0 to 100 scale.
Output price
DeepSeek V4 Flash 0731 $0.28 against GLM 5.2 $4.40 per million output tokens, 15.7x apart.
Context window
Both take 1.05M tokens.

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 Flash 0731BarsGLM 5.2Ties
Coding75
Intelligence20
Agentic work13
Reasoning72
Math22
Knowledge15
Web design08
General13
All shared boards2128
CodingLike-for-likeSix of these boards feed the Vibe Coding Index75
Coding benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
LMArena WebDevNot measured1604Only GLM 5.2 measured
AA Coding Index69.168.8DeepSeek V4 Flash 0731
LiveCodeBench (Vals AI)87.3%69.5%DeepSeek V4 Flash 0731
SWE-bench Verified (Vals AI)88.8%82.8%DeepSeek V4 Flash 0731
Terminal-Bench 4 (Vals AI)18.7%Not measuredOnly DeepSeek V4 Flash 0731 measured
Vibe Code Bench (Vals AI)74.7%64.0%DeepSeek V4 Flash 0731
Code migration (Vals AI)38.6%37.9%DeepSeek V4 Flash 0731
ProgramBench (Vals AI)0.0%0.5%GLM 5.2
IOI (Vals AI)32.7%Not measuredOnly DeepSeek V4 Flash 0731 measured
Terminal-Bench 2.1 (Vals AI)67.0%67.8%GLM 5.2
SkillsBench (Vals AI)50.7%45.1%DeepSeek V4 Flash 0731
CyberBench (Vals AI)72.1%Not measuredOnly DeepSeek V4 Flash 0731 measured
SciCode50.3%51.2%GLM 5.2
Terminal-Bench Hard (Artificial Analysis run)Not measured50.8%Only GLM 5.2 measured
FrontierCode18.8%24.5%GLM 5.2
WeirdML63.0%70.1%GLM 5.2
ALE-Bench1306.11047.0DeepSeek V4 Flash 0731
LMArena Text: codingNot measured1487Only GLM 5.2 measured
LMArena Text: software and ITNot measured1486Only GLM 5.2 measured
LMArena WebDev: ReactNot measured1617Only GLM 5.2 measured
LMArena WebDev: HTMLNot measured1565Only GLM 5.2 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 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
AA Intelligence Index34.333.7DeepSeek V4 Flash 0731
Epoch Capabilities Index154.8151.9DeepSeek V4 Flash 0731
Agentic workLike-for-likePublic boards, shown as published, kept out of the index13
Agentic work benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
AA Agentic Index41.038.4DeepSeek V4 Flash 0731
Design Arena: agents building web appsNot measured1222Only GLM 5.2 measured
Design Arena: agents building full-stack appsNot measured1241Only GLM 5.2 measured
Design Arena: agents building mobile appsNot measured1160Only GLM 5.2 measured
Design Arena: agents building Android appsNot measured1178Only GLM 5.2 measured
Design Arena: agents building Godot gamesNot measured1142Only GLM 5.2 measured
Design Arena: agents building gamesNot measured1206Only GLM 5.2 measured
Design Arena: HTML slidesNot measured1175Only GLM 5.2 measured
Design Arena: PowerPoint slidesNot measured1195Only GLM 5.2 measured
SREBench (Vals AI)Not measured0.0%Only GLM 5.2 measured
Finance Agent (Vals AI)49.5%49.7%GLM 5.2
Tax Agent Bench (Vals AI)28.1%Not measuredOnly DeepSeek V4 Flash 0731 measured
Legal Research Bench (Vals AI)30.3%31.3%GLM 5.2
Excel Modeling Benchmark (Vals AI)57.0%61.5%GLM 5.2
BioMysteryBench (Vals AI)64.4%Not measuredOnly DeepSeek V4 Flash 0731 measured
Terminal-Bench-Science (Vals AI)4.3%Not measuredOnly DeepSeek V4 Flash 0731 measured
τ²-Bench TelecomNot measured99.1%Only GLM 5.2 measured
APEX-AgentsNot measured45.2%Only GLM 5.2 measured
PostTrainBenchNot measured31.7%Only GLM 5.2 measured
Vending-Bench 2Not measured$8,314Only GLM 5.2 measured
Neither model has a result on the other 21 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index72
Reasoning benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
GPQA Diamond89.9%85.6%DeepSeek V4 Flash 0731
GPQA Diamond (Artificial Analysis run)90.8%89.5%DeepSeek V4 Flash 0731
Humanity's Last Exam (Artificial Analysis run)38.6%41.1%GLM 5.2
GPQA Diamond (Epoch AI run)91.0%91.9%GLM 5.2
ARC-AGI89.0%77.0%DeepSeek V4 Flash 0731
ARC-AGI-261.4%22.8%DeepSeek V4 Flash 0731
SimpleBench61.1%58.8%DeepSeek V4 Flash 0731
Chess puzzles32.0%22.0%DeepSeek V4 Flash 0731
Mystery game puzzles34.0%19.0%DeepSeek V4 Flash 0731
EBR-benchNot measured9.5%Only GLM 5.2 measured
LMArena Text: hard promptsNot measured1481Only GLM 5.2 measured
LMArena Text: expertNot measured1487Only GLM 5.2 measured
LMArena Text: hard prompts in EnglishNot measured1481Only GLM 5.2 measured
Kagi LLM BenchmarkNot measured62.6%Only GLM 5.2 measured
Neither model has a result on the other 8 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index22
Math benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
OTIS Mock AIME94.4%86.4%DeepSeek V4 Flash 0731
FrontierMath, tiers 1 to 357.5%59.2%GLM 5.2
FrontierMath, tier 424.4%29.3%GLM 5.2
ProofBench56.0%35.0%DeepSeek V4 Flash 0731
LMArena Text: mathNot measured1483Only GLM 5.2 measured
LMArena Text: mathematical workNot measured1489Only GLM 5.2 measured
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index15
Knowledge benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
MMLU Pro86.2%86.7%GLM 5.2
LegalBench (Vals AI)77.7%84.1%GLM 5.2
CorpFin (Vals AI)61.9%66.1%GLM 5.2
TaxEval (Vals AI)70.7%73.3%GLM 5.2
MedCode (Vals AI)41.4%40.8%DeepSeek V4 Flash 0731
SimpleQA Verified33.6%34.2%GLM 5.2
LMArena Text: scienceNot measured1489Only GLM 5.2 measured
LMArena Text: medicine and healthcareNot measured1475Only GLM 5.2 measured
LMArena Text: legal and governmentNot measured1486Only GLM 5.2 measured
LMArena Text: business and financeNot measured1451Only GLM 5.2 measured
Neither model has a result on the other 8 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index08
Web design benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
Design Arena: websites12381285GLM 5.2
Design Arena: UI components12351296GLM 5.2
Design Arena: game dev12081268GLM 5.2
Design Arena: data viz11801285GLM 5.2
Design Arena: 3D12091288GLM 5.2
Design Arena: all code categories12301288GLM 5.2
Design Arena: SVG11851214GLM 5.2
Design Arena: ASCII art10901190GLM 5.2
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index13
General benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
MedScribe (Vals AI)80.4%83.5%GLM 5.2
IFBenchNot measured73.3%Only GLM 5.2 measured
Long context reasoning (AA-LCR)79.7%78.3%DeepSeek V4 Flash 0731
DTBench90.9%93.6%GLM 5.2
LMCA41.7%45.8%GLM 5.2
Surface Evolver BenchNot measured55.6%Only GLM 5.2 measured
LMArena TextNot measured1470Only GLM 5.2 measured
LMArena Text: instruction followingNot measured1466Only GLM 5.2 measured
LMArena Text: creative writingNot measured1463Only GLM 5.2 measured
LMArena Text: EnglishNot measured1475Only GLM 5.2 measured
LMArena Text: non-EnglishNot measured1459Only GLM 5.2 measured
LMArena Text: ChineseNot measured1522Only GLM 5.2 measured
LMArena Text: GermanNot measured1470Only GLM 5.2 measured
LMArena Text: JapaneseNot measured1450Only GLM 5.2 measured
LMArena Text: KoreanNot measured1440Only GLM 5.2 measured
LMArena Text: FrenchNot measured1479Only GLM 5.2 measured
LMArena Text: SpanishNot measured1478Only GLM 5.2 measured
LMArena Text: RussianNot measured1468Only GLM 5.2 measured
LMArena Text: PolishNot measured1456Only GLM 5.2 measured
LMArena Text: multi-turnNot measured1469Only GLM 5.2 measured
LMArena Text: longer queriesNot measured1480Only GLM 5.2 measured
LMArena Text: writing and languageNot measured1463Only GLM 5.2 measured
LMArena Text: entertainment and mediaNot measured1450Only GLM 5.2 measured
Neither model has a result on the other 23 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V4 Flash 0731 and GLM 5.2
BenchmarkDeepSeek V4 Flash 0731GLM 5.2Leader
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 Flash 0731$0.0003Fits in one request
GLM 5.2$0.0036Fits in one request

DeepSeek V4 Flash 0731 costs less

Repository review

50K input + 3K output tokens

DeepSeek V4 Flash 0731$0.0078Fits in one request
GLM 5.2$0.083Fits in one request

DeepSeek V4 Flash 0731 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V4 Flash 0731$0.031Fits in one request
GLM 5.2$0.324Fits in one request

DeepSeek V4 Flash 0731 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 Flash 0731 and GLM 5.2
SpecificationDeepSeek V4 Flash 0731Z.ai GLM 5.2
MakerDeepSeekZ.ai
WeightsOpen weightsOpen weights
ReleasedJuly 31, 2026June 16, 2026
Context window1.05M1.05M
Input $/Mtok$0.14$1.40
Output $/Mtok$0.28$4.40
Coding index51.861.1
Independent measurements behind it44
Cost per solved task$0.01$0.86
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V4 Flash 0731 better than GLM 5.2 for coding?

Pick GLM 5.2 for coding: it scores 61.1 on the Vibe Coding Index against 51.8 for DeepSeek V4 Flash 0731. Pick DeepSeek V4 Flash 0731 for output tokens 15.7x cheaper ($0.28 vs $4.40 per million) and a lower cost per solved task ($0.01 against $0.86).

Which is cheaper, DeepSeek V4 Flash 0731 or GLM 5.2?

DeepSeek V4 Flash 0731 costs $0.14 per million input tokens and $0.28 per million output tokens; GLM 5.2 costs $1.40 and $4.40. Per solved SWE-bench Verified task, DeepSeek V4 Flash 0731 costs $0.01 and GLM 5.2 $0.86.

Which has the bigger context window, DeepSeek V4 Flash 0731 or GLM 5.2?

Neither: both take 1.05M tokens.

How many benchmarks is the DeepSeek V4 Flash 0731 vs GLM 5.2 comparison based on?

49 benchmarks have a result for both models as of October 21, 2026. DeepSeek V4 Flash 0731's coding index rests on 4 independent benchmarks and GLM 5.2's on 4. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/deepseek-v4-flash-0731-vs-glm-5-2" title="DeepSeek V4 Flash 0731 vs GLM 5.2, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/deepseek-v4-flash-0731-vs-glm-5-2">LLMBenchmark.ai, DeepSeek V4 Flash 0731 vs GLM 5.2</a></p>