LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

GLM 5.2 vs GLM 5.3

Updated October 21, 2026. 82 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace GLM 5.2 Replace GLM 5.3
GLM 5.2Z.ai
Coding
57.7
Overall
33.7
Output $/M
$7.00
10
GLM 5.3Z.ai
Coding
73.1
Overall
44.8
Output $/M
$1.40
71
vs boards won of 82 shared, 1 tied

Decision reading

Pick GLM 5.3 for coding: it scores 73.1 on the Vibe Coding Index against 57.7 for GLM 5.2.

Pick GLM 5.2 when what matters is Excel Modeling Benchmark (Vals AI) or Vending-Bench 2, where it leads.

Model AGLM 5.2Z.ai57.7/1001 independent benchmark$0.86 per solved taskModel BGLM 5.3Z.ai73.1/1002 independent benchmarks$0.36 per solved task
Shared results
82
GLM 5.2 only
15
GLM 5.3 only
18
Like-for-like categories
8 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

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
GLM 5.2 57.7 from 1 benchmark, GLM 5.3 73.1 from 2 benchmarks.
Widest category
Intelligence, 8.1 points apart on average over 2 shared benchmarks.
Widest single gap
Legal Research Bench (Vals AI): GLM 5.2 31.3% against GLM 5.3 49.0%, 17.8 points apart on a 0 to 100 scale.
Output price
GLM 5.2 $7.00 against GLM 5.3 $1.40 per million output tokens, 5.0x 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
CategoryGLM 5.2BarsGLM 5.3Ties
Coding017
Intelligence02
Agentic work28
Reasoning18
Math051 tie
Knowledge27
Web design17
General417
All shared boards10711 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index017
Coding benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
LiveBench codingNot measured82.8%Only GLM 5.3 measured
LMArena WebDev16041621GLM 5.3
AA Coding Index68.874.8GLM 5.3
LiveCodeBench (Vals AI)69.5%80.5%GLM 5.3
SWE-bench Verified (Vals AI)82.8%95.4%GLM 5.3
Terminal-Bench 4 (Vals AI)Not measured38.9%Only GLM 5.3 measured
Vibe Code Bench (Vals AI)64.0%78.1%GLM 5.3
Code migration (Vals AI)37.9%44.2%GLM 5.3
ProgramBench (Vals AI)0.5%1.5%GLM 5.3
IOI (Vals AI)Not measured68.4%Only GLM 5.3 measured
Terminal-Bench 2.1 (Vals AI)67.8%71.5%GLM 5.3
Vibe Code Bench 1-100 (Vals AI)Not measured20.0%Only GLM 5.3 measured
SkillsBench (Vals AI)45.1%47.5%GLM 5.3
CyberBench (Vals AI)Not measured72.1%Only GLM 5.3 measured
SciCode51.2%59.0%GLM 5.3
Terminal-Bench Hard (Artificial Analysis run)50.8%Not measuredOnly GLM 5.2 measured
FrontierCode24.5%40.1%GLM 5.3
FrontierSWENot measured30.2%Only GLM 5.3 measured
WeirdML70.1%75.4%GLM 5.3
ALE-Bench1047.01317.4GLM 5.3
LMArena Text: coding14871497GLM 5.3
LMArena Text: software and IT14861493GLM 5.3
LMArena WebDev: React16171641GLM 5.3
LMArena WebDev: HTML15651595GLM 5.3
Neither model has a result on the other 14 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
AA Intelligence Index33.744.8GLM 5.3
Epoch Capabilities Index151.9156.0GLM 5.3
Agentic workLike-for-likePublic boards, shown as published, kept out of the index28
Agentic work benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
AA Agentic Index38.453.1GLM 5.3
Design Arena: agents building web apps1222Not measuredOnly GLM 5.2 measured
Design Arena: agents building full-stack apps1241Not measuredOnly GLM 5.2 measured
Design Arena: agents building mobile apps11601194GLM 5.3
Design Arena: agents building Android apps1184Not measuredOnly GLM 5.2 measured
Design Arena: agents building Godot games1142Not measuredOnly GLM 5.2 measured
Design Arena: agents building games1206Not measuredOnly GLM 5.2 measured
Design Arena: HTML slides11751189GLM 5.3
Design Arena: PowerPoint slides11951255GLM 5.3
LiveBench researchNot measured73.2%Only GLM 5.3 measured
LiveBench browserNot measured0.0%Only GLM 5.3 measured
LiveBench long agenticNot measured35.9%Only GLM 5.3 measured
SREBench (Vals AI)0.0%9.2%GLM 5.3
Finance Agent (Vals AI)49.7%55.8%GLM 5.3
Tax Agent Bench (Vals AI)Not measured39.7%Only GLM 5.3 measured
Legal Research Bench (Vals AI)31.3%49.0%GLM 5.3
Harvey's Legal Agent Benchmark (Vals AI)Not measured14.6%Only GLM 5.3 measured
Excel Modeling Benchmark (Vals AI)61.5%56.3%GLM 5.2
Terminal-Bench-Science (Vals AI)Not measured5.7%Only GLM 5.3 measured
τ²-Bench Telecom99.1%Not measuredOnly GLM 5.2 measured
APEX-Agents45.2%56.6%GLM 5.3
PostTrainBench31.7%Not measuredOnly GLM 5.2 measured
Vending-Bench 2$8,314$8,164GLM 5.2
Neither model has a result on the other 16 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index18
Reasoning benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
GPQA Diamond85.6%88.1%GLM 5.3
LiveBench reasoningNot measured85.8%Only GLM 5.3 measured
MysteryMechanism (Vals AI)Not measured23.0%Only GLM 5.3 measured
GPQA Diamond (Artificial Analysis run)89.5%91.7%GLM 5.3
Humanity's Last Exam (Artificial Analysis run)41.1%42.3%GLM 5.3
GPQA Diamond (Epoch AI run)91.9%90.9%GLM 5.2
ARC-AGI77.0%Not measuredOnly GLM 5.2 measured
ARC-AGI-222.8%Not measuredOnly GLM 5.2 measured
SimpleBench58.8%Not measuredOnly GLM 5.2 measured
Chess puzzles22.0%37.0%GLM 5.3
Mystery game puzzles19.0%33.0%GLM 5.3
EBR-bench9.5%Not measuredOnly GLM 5.2 measured
LMArena Text: hard prompts14811490GLM 5.3
LMArena Text: expert14871514GLM 5.3
LMArena Text: hard prompts in English14811496GLM 5.3
Neither model has a result on the other 6 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index05
Math benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
OTIS Mock AIME86.4%91.1%GLM 5.3
FrontierMath, tiers 1 to 359.2%68.8%GLM 5.3
FrontierMath, tier 429.3%29.3%Tie
ProofBench35.0%49.0%GLM 5.3
LMArena Text: math14831495GLM 5.3
LMArena Text: mathematical work14891497GLM 5.3
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index27
Knowledge benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
MMLU Pro86.7%86.8%GLM 5.3
LegalBench (Vals AI)84.1%84.8%GLM 5.3
CorpFin (Vals AI)66.1%Not measuredOnly GLM 5.2 measured
TaxEval (Vals AI)73.3%72.4%GLM 5.2
MedCode (Vals AI)40.8%42.9%GLM 5.3
SimpleQA Verified34.2%41.0%GLM 5.3
LMArena Text: science14891498GLM 5.3
LMArena Text: medicine and healthcare14751481GLM 5.3
LMArena Text: legal and government14861473GLM 5.2
LMArena Text: business and finance14511459GLM 5.3
Neither model has a result on the other 7 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index17
Web design benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
Design Arena: websites12851295GLM 5.3
Design Arena: UI components12961314GLM 5.3
Design Arena: game dev12681311GLM 5.3
Design Arena: data viz12851260GLM 5.2
Design Arena: 3D12881332GLM 5.3
Design Arena: all code categories12881304GLM 5.3
Design Arena: SVG12141292GLM 5.3
Design Arena: ASCII art11901213GLM 5.3
LiveBench designNot measured77.3%Only GLM 5.3 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index417
General benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
Vals IndexNot measured59.7%Only GLM 5.3 measured
LiveBench data analysisNot measured70.2%Only GLM 5.3 measured
Public Benefits Bench (Vals AI)Not measured68.5%Only GLM 5.3 measured
MedScribe (Vals AI)83.5%88.8%GLM 5.3
IFBench73.3%Not measuredOnly GLM 5.2 measured
Long context reasoning (AA-LCR)78.3%79.7%GLM 5.3
DTBench93.6%87.7%GLM 5.2
LMCA45.8%55.5%GLM 5.3
Surface Evolver Bench55.6%Not measuredOnly GLM 5.2 measured
LMArena Text14701474GLM 5.3
LMArena Text: instruction following14661479GLM 5.3
LMArena Text: creative writing14631458GLM 5.2
LMArena Text: English14751482GLM 5.3
LMArena Text: non-English14591459GLM 5.3
LMArena Text: Chinese15221532GLM 5.3
LMArena Text: German14701488GLM 5.3
LMArena Text: Japanese14501460GLM 5.3
LMArena Text: Korean14401460GLM 5.3
LMArena Text: French14791503GLM 5.3
LMArena Text: Spanish14781462GLM 5.2
LMArena Text: Russian14681469GLM 5.3
LMArena Text: Polish14561466GLM 5.3
LMArena Text: multi-turn14691472GLM 5.3
LMArena Text: longer queries14801483GLM 5.3
LMArena Text: writing and language14631469GLM 5.3
LMArena Text: entertainment and media14501447GLM 5.2
Neither model has a result on the other 20 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GLM 5.2 and GLM 5.3
BenchmarkGLM 5.2GLM 5.3Leader
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

GLM 5.2$0.0035Fits in one request
GLM 5.3$0.0008Fits in one request

GLM 5.3 costs less

Repository review

50K input + 3K output tokens

GLM 5.2$0.023Fits in one request
GLM 5.3$0.0081Fits in one request

GLM 5.3 costs less

Agent loop

200K input + 10K output tokens

GLM 5.2$0.077Fits in one request
GLM 5.3$0.030Fits in one request

GLM 5.3 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 GLM 5.2 and GLM 5.3
SpecificationZ.ai GLM 5.2Z.ai GLM 5.3
MakerZ.aiZ.ai
WeightsOpen weightsOpen weights
ReleasedJune 16, 2026August 18, 2026
Context window1.05M1.05M
Input $/Mtok$0.0341$0.079
Output $/Mtok$7.00$1.40
Coding index57.773.1
Independent measurements behind it12
Cost per solved task$0.86$0.36
SourceOpenRouter listingOpenRouter listing

Questions

Is GLM 5.2 better than GLM 5.3 for coding?

Pick GLM 5.3 for coding: it scores 73.1 on the Vibe Coding Index against 57.7 for GLM 5.2. Pick GLM 5.2 when what matters is Excel Modeling Benchmark (Vals AI) or Vending-Bench 2, where it leads.

Which is cheaper, GLM 5.2 or GLM 5.3?

GLM 5.2 costs $0.0341 per million input tokens and $7.00 per million output tokens; GLM 5.3 costs $0.079 and $1.40. Per solved SWE-bench Verified task, GLM 5.2 costs $0.86 and GLM 5.3 $0.36.

Which has the bigger context window, GLM 5.2 or GLM 5.3?

Neither: both take 1.05M tokens.

How many benchmarks is the GLM 5.2 vs GLM 5.3 comparison based on?

82 benchmarks have a result for both models as of October 21, 2026. GLM 5.2's coding index rests on 1 independent benchmark and GLM 5.3's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next