Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 vs GLM 4.7

GLM 4.7 leads DeepSeek V3.2 on coding, 55.1 to 30.0 on the Vibe Coding Index, and DeepSeek V3.2 costs 2.5x less per output token ($0.88 vs $2.20 per million). Updated October 22, 2026; 67 benchmarks have a result for both.

Replace DeepSeek V3.2 Replace GLM 4.7
DeepSeek V3.2DeepSeek
Coding
30.0
Overall
37.9
Output $/M
$0.88
15
GLM 4.7Z.ai
Coding
55.1
Overall
56.9
Output $/M
$2.20
52
vs boards won of 67 shared

DeepSeek V3.2 or GLM 4.7: which to pick

Pick GLM 4.7 for coding: it scores 55.1 on the Vibe Coding Index against 30.0 for DeepSeek V3.2, though on 2 benchmarks against 3.

Pick DeepSeek V3.2 for output tokens 2.5x cheaper ($0.88 vs $2.20 per million) and work measured by Terminal-Bench Hard (Artificial Analysis run) or Terminal-Bench, where it leads.

Model ADeepSeek V3.2DeepSeek30.0/1003 independent benchmarks$0.99 per solved taskModel BGLM 4.7Z.ai55.1/1002 independent benchmarks$0.29 per solved task
Shared results
67
DeepSeek V3.2 only
8
GLM 4.7 only
14
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 V3.2 30.0 from 3 benchmarks, GLM 4.7 55.1 from 2 benchmarks.
Widest category
Web design, 7.8 points apart on average over 8 shared benchmarks.
Widest single gap
Design Arena: SVG: DeepSeek V3.2 1032 against GLM 4.7 1142, 13.8 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 $0.88 against GLM 4.7 $2.20 per million output tokens, 2.5x apart.
Context window
DeepSeek V3.2 takes 164K tokens, GLM 4.7 205K.

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 V3.2BarsGLM 4.7Ties
Coding29
Intelligence11
Agentic work02
Reasoning24
Math24
Knowledge56
Web design08
General318
All shared boards1552
CodingLike-for-likeSix of these boards feed the Vibe Coding Index29
Coding benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
SWE-bench Verified70.0%Not measuredOnly DeepSeek V3.2 measured
SWE-bench Multilingual59.0%Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev13621435GLM 4.7
AA Coding Index44.245.3GLM 4.7
LiveCodeBench (Vals AI)80.7%82.2%GLM 4.7
SWE-bench Verified (Vals AI)67.6%69.4%GLM 4.7
Vibe Code Bench (Vals AI)5.1%Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Artificial Analysis run)86.2%89.4%GLM 4.7
Terminal-Bench Hard (Artificial Analysis run)35.6%31.8%DeepSeek V3.2
Terminal-Bench39.6%33.4%DeepSeek V3.2
ALE-BenchNot measured399.5Only GLM 4.7 measured
LMArena Text: coding14531454GLM 4.7
LMArena Text: software and IT14491453GLM 4.7
LMArena WebDev: React13501420GLM 4.7
LMArena WebDev: HTML13711447GLM 4.7
Neither model has a result on the other 24 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index11
Intelligence benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
AA Intelligence Index21.522.2GLM 4.7
Epoch Capabilities Index146.1143.6DeepSeek V3.2
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
Agentic work benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
Design Arena: agents building full-stack appsNot measured1050Only GLM 4.7 measured
Design Arena: agents building mobile appsNot measured1105Only GLM 4.7 measured
Design Arena: agents building Android appsNot measured1112Only GLM 4.7 measured
Design Arena: agents building Godot gamesNot measured1058Only GLM 4.7 measured
τ²-Bench Telecom90.6%95.9%GLM 4.7
APEX-Agents21.3%Not measuredOnly DeepSeek V3.2 measured
Vending-Bench 2$1,034$2,377GLM 4.7
BrowseComp-PlusNot measured64.7%Only GLM 4.7 measured
Neither model has a result on the other 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index24
Reasoning benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
GPQA Diamond80.3%80.0%DeepSeek V3.2
GPQA Diamond (Artificial Analysis run)84.0%85.9%GLM 4.7
Humanity's Last Exam (Artificial Analysis run)24.6%27.4%GLM 4.7
GPQA Diamond (Epoch AI run)Not measured83.3%Only GLM 4.7 measured
ARC-AGI57.0%Not measuredOnly DeepSeek V3.2 measured
ARC-AGI-24.0%Not measuredOnly DeepSeek V3.2 measured
SimpleBenchNot measured47.7%Only GLM 4.7 measured
Chess puzzlesNot measured21.0%Only GLM 4.7 measured
LMArena Text: hard prompts14341443GLM 4.7
LMArena Text: expert14361424DeepSeek V3.2
LMArena Text: hard prompts in English14461452GLM 4.7
Neither model has a result on the other 11 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index24
Math benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
AIME84.6%93.3%GLM 4.7
MGSM (Vals AI)86.0%88.2%GLM 4.7
AA Math Index92.095.0GLM 4.7
AIME 2025 (Artificial Analysis run)92.0%95.0%GLM 4.7
OTIS Mock AIMENot measured83.3%Only GLM 4.7 measured
ProofBenchNot measured6.0%Only GLM 4.7 measured
LMArena Text: math14351424DeepSeek V3.2
LMArena Text: mathematical work14401411DeepSeek V3.2
Neither model has a result on the other 5 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index56
Knowledge benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
MMLU Pro84.9%82.7%DeepSeek V3.2
LegalBench (Vals AI)76.1%83.4%GLM 4.7
CaseLaw (Vals AI)55.4%54.9%DeepSeek V3.2
CorpFin (Vals AI)51.0%46.4%DeepSeek V3.2
TaxEval (Vals AI)68.2%68.8%GLM 4.7
MedQA (Vals AI)93.9%93.7%DeepSeek V3.2
MedCode (Vals AI)Not measured32.8%Only GLM 4.7 measured
MMLU Pro (Artificial Analysis run)86.2%85.6%DeepSeek V3.2
SimpleQA VerifiedNot measured32.2%Only GLM 4.7 measured
LMArena Text: science14461457GLM 4.7
LMArena Text: medicine and healthcare14401469GLM 4.7
LMArena Text: legal and government14251444GLM 4.7
LMArena Text: business and finance14101429GLM 4.7
Neither model has a result on the other 5 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index08
Web design benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
Design Arena: websites11751225GLM 4.7
Design Arena: UI components11461196GLM 4.7
Design Arena: game dev11311188GLM 4.7
Design Arena: data viz11581195GLM 4.7
Design Arena: 3D11321190GLM 4.7
Design Arena: all code categories11631216GLM 4.7
Design Arena: SVG10321142GLM 4.7
Design Arena: ASCII art10831165GLM 4.7
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index318
General benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
MedScribe (Vals AI)Not measured68.6%Only GLM 4.7 measured
IFBench60.7%67.9%GLM 4.7
Long context reasoning (AA-LCR)73.3%71.0%DeepSeek V3.2
DTBench85.6%Not measuredOnly DeepSeek V3.2 measured
LMCA28.8%Not measuredOnly DeepSeek V3.2 measured
CL-bench12.4%15.9%GLM 4.7
CL-bench Life7.4%10.9%GLM 4.7
LMArena Text14251435GLM 4.7
LMArena Text: instruction following14131412DeepSeek V3.2
LMArena Text: creative writing14001401GLM 4.7
LMArena Text: English14391453GLM 4.7
LMArena Text: non-English14091417GLM 4.7
LMArena Text: Chinese14551495GLM 4.7
LMArena Text: German14131426GLM 4.7
LMArena Text: Japanese13741439GLM 4.7
LMArena Text: Korean13701399GLM 4.7
LMArena Text: French14291432GLM 4.7
LMArena Text: Spanish14241434GLM 4.7
LMArena Text: Russian14241423DeepSeek V3.2
LMArena Text: Polish14191435GLM 4.7
LMArena Text: multi-turn14271446GLM 4.7
LMArena Text: longer queries14281432GLM 4.7
LMArena Text: writing and language14071411GLM 4.7
LMArena Text: entertainment and media13951400GLM 4.7
Neither model has a result on the other 22 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V3.2 and GLM 4.7
BenchmarkDeepSeek V3.2GLM 4.7Leader
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 V3.2$0.0007Fits in one request
GLM 4.7$0.0017Fits in one request

DeepSeek V3.2 costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2$0.017Fits in one request
GLM 4.7$0.037Fits in one request

DeepSeek V3.2 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2$0.067Exceeds the context window
GLM 4.7$0.142Exceeds the context window

DeepSeek V3.2 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 V3.2 and GLM 4.7
SpecificationDeepSeek V3.2Z.ai GLM 4.7
MakerDeepSeekZ.ai
WeightsOpen weightsOpen weights
ReleasedDecember 1, 2025December 22, 2025
Context window164K205K
Input $/Mtok$0.29$0.60
Output $/Mtok$0.88$2.20
Coding index30.055.1
Independent measurements behind it32
Cost per solved task$0.99$0.29
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 better than GLM 4.7 for coding?

Pick GLM 4.7 for coding: it scores 55.1 on the Vibe Coding Index against 30.0 for DeepSeek V3.2, though on 2 benchmarks against 3. Pick DeepSeek V3.2 for output tokens 2.5x cheaper ($0.88 vs $2.20 per million) and work measured by Terminal-Bench Hard (Artificial Analysis run) or Terminal-Bench, where it leads.

Which is cheaper, DeepSeek V3.2 or GLM 4.7?

DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; GLM 4.7 costs $0.60 and $2.20. Per solved SWE-bench Verified task, DeepSeek V3.2 costs $0.99 and GLM 4.7 $0.29.

Which has the bigger context window, DeepSeek V3.2 or GLM 4.7?

GLM 4.7, with 205K tokens against 164K.

How many benchmarks is the DeepSeek V3.2 vs GLM 4.7 comparison based on?

67 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and GLM 4.7's on 2. 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-v3-2-vs-glm-4-7" title="DeepSeek V3.2 vs GLM 4.7, 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-v3-2-vs-glm-4-7">LLMBenchmark.ai, DeepSeek V3.2 vs GLM 4.7</a></p>