LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

GLM 5.2 vs GLM 5.3 Flash

Updated October 21, 2026. 76 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 Flash
GLM 5.2Z.ai
Coding
57.7
Overall
33.7
Output $/M
$7.00
33
GLM 5.3 FlashZ.ai
Coding
71.0
Overall
41.8
Output $/M
$0.50
43
vs boards won of 76 shared

Decision reading

Pick GLM 5.3 Flash for coding: it scores 71.0 on the Vibe Coding Index against 57.7 for GLM 5.2.

Pick GLM 5.2 when what matters is Vibe Code Bench (Vals AI) or Code migration (Vals AI), where it leads.

Model AGLM 5.2Z.ai57.7/1001 independent benchmark$0.86 per solved taskModel BGLM 5.3 FlashZ.ai71.0/1002 independent benchmarks$0.02 per solved task
Shared results
76
GLM 5.2 only
21
GLM 5.3 Flash only
26
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 Flash 71.0 from 2 benchmarks.
Widest category
Agentic work, 8.1 points apart on average over 6 shared benchmarks.
Widest single gap
ARC-AGI-2: GLM 5.2 22.8% against GLM 5.3 Flash 65.8%, 43.1 points apart on a 0 to 100 scale.
Output price
GLM 5.2 $7.00 against GLM 5.3 Flash $0.50 per million output tokens, 14.0x 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 5.3 FlashTies
Coding610
Intelligence02
Agentic work15
Reasoning47
Math33
Knowledge52
Web design35
General119
All shared boards3343
CodingLike-for-likeFour of these boards feed the Vibe Coding Index610
Coding benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
LiveBench codingNot measured80.6%Only GLM 5.3 Flash measured
LMArena WebDev16041607GLM 5.3 Flash
AA Coding Index68.871.5GLM 5.3 Flash
LiveCodeBench (Vals AI)69.5%80.5%GLM 5.3 Flash
SWE-bench Verified (Vals AI)82.8%92.0%GLM 5.3 Flash
Terminal-Bench 4 (Vals AI)Not measured37.4%Only GLM 5.3 Flash measured
Vibe Code Bench (Vals AI)64.0%30.8%GLM 5.2
Code migration (Vals AI)37.9%20.5%GLM 5.2
ProgramBench (Vals AI)0.5%0.0%GLM 5.2
IOI (Vals AI)Not measured52.5%Only GLM 5.3 Flash measured
Terminal-Bench 2.1 (Vals AI)67.8%62.9%GLM 5.2
Vibe Code Bench 1-100 (Vals AI)Not measured16.0%Only GLM 5.3 Flash measured
SkillsBench (Vals AI)45.1%40.2%GLM 5.2
CyberBench (Vals AI)Not measured67.9%Only GLM 5.3 Flash measured
SciCode51.2%51.6%GLM 5.3 Flash
Terminal-Bench Hard (Artificial Analysis run)50.8%Not measuredOnly GLM 5.2 measured
FrontierCode24.5%31.8%GLM 5.3 Flash
FrontierSWENot measured18.1%Only GLM 5.3 Flash measured
WeirdML70.1%Not measuredOnly GLM 5.2 measured
ALE-Bench1047.0303.6GLM 5.2
LMArena Text: coding14871507GLM 5.3 Flash
LMArena Text: software and IT14861500GLM 5.3 Flash
LMArena WebDev: React16171624GLM 5.3 Flash
LMArena WebDev: HTML15651577GLM 5.3 Flash
LMArena WebDev: from an imageNot measured1578Only GLM 5.3 Flash measured
Neither model has a result on the other 13 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
AA Intelligence Index33.741.8GLM 5.3 Flash
Epoch Capabilities Index151.9153.1GLM 5.3 Flash
Agentic workLike-for-likePublic boards, shown as published, kept out of the index15
Agentic work benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
AA Agentic Index38.450.9GLM 5.3 Flash
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 apps1160Not measuredOnly GLM 5.2 measured
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 slides1175Not measuredOnly GLM 5.2 measured
Design Arena: PowerPoint slides1195Not measuredOnly GLM 5.2 measured
LiveBench researchNot measured70.1%Only GLM 5.3 Flash measured
LiveBench browserNot measured84.0%Only GLM 5.3 Flash measured
LiveBench long agenticNot measured28.4%Only GLM 5.3 Flash measured
SREBench (Vals AI)0.0%0.8%GLM 5.3 Flash
Finance Agent (Vals AI)49.7%57.9%GLM 5.3 Flash
Tax Agent Bench (Vals AI)Not measured31.6%Only GLM 5.3 Flash measured
Legal Research Bench (Vals AI)31.3%45.2%GLM 5.3 Flash
Harvey's Legal Agent Benchmark (Vals AI)Not measured10.4%Only GLM 5.3 Flash measured
Excel Modeling Benchmark (Vals AI)61.5%55.9%GLM 5.2
Terminal-Bench-Science (Vals AI)Not measured2.9%Only GLM 5.3 Flash measured
τ²-Bench Telecom99.1%Not measuredOnly GLM 5.2 measured
APEX-Agents45.2%52.8%GLM 5.3 Flash
PostTrainBench31.7%Not measuredOnly GLM 5.2 measured
Vending-Bench 2$8,314Not measuredOnly GLM 5.2 measured
Neither model has a result on the other 16 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index47
Reasoning benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
GPQA Diamond85.6%86.4%GLM 5.3 Flash
LiveBench reasoningNot measured77.6%Only GLM 5.3 Flash measured
MMMU Pro (Vals AI)Not measured86.0%Only GLM 5.3 Flash measured
GPQA Diamond (Artificial Analysis run)89.5%91.2%GLM 5.3 Flash
Humanity's Last Exam (Artificial Analysis run)41.1%39.9%GLM 5.2
GPQA Diamond (Epoch AI run)91.9%90.2%GLM 5.2
ARC-AGI77.0%91.0%GLM 5.3 Flash
ARC-AGI-222.8%65.8%GLM 5.3 Flash
SimpleBench58.8%Not measuredOnly GLM 5.2 measured
Chess puzzles22.0%21.0%GLM 5.2
Mystery game puzzles19.0%8.0%GLM 5.2
EBR-bench9.5%Not measuredOnly GLM 5.2 measured
LMArena Text: hard prompts14811490GLM 5.3 Flash
LMArena Text: expert14871511GLM 5.3 Flash
LMArena Text: hard prompts in English14811488GLM 5.3 Flash
LMArena Vision: diagramsNot measured1317Only GLM 5.3 Flash measured
LMArena Vision: homeworkNot measured1318Only GLM 5.3 Flash measured
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index33
Math benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
OTIS Mock AIME86.4%93.9%GLM 5.3 Flash
FrontierMath, tiers 1 to 359.2%55.8%GLM 5.2
FrontierMath, tier 429.3%17.1%GLM 5.2
ProofBench35.0%21.0%GLM 5.2
LMArena Text: math14831503GLM 5.3 Flash
LMArena Text: mathematical work14891497GLM 5.3 Flash
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index52
Knowledge benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
MMLU Pro86.7%86.1%GLM 5.2
LegalBench (Vals AI)84.1%83.9%GLM 5.2
CorpFin (Vals AI)66.1%Not measuredOnly GLM 5.2 measured
TaxEval (Vals AI)73.3%75.6%GLM 5.3 Flash
MedCode (Vals AI)40.8%Not measuredOnly GLM 5.2 measured
SimpleQA Verified34.2%Not measuredOnly GLM 5.2 measured
LMArena Text: science14891484GLM 5.2
LMArena Text: medicine and healthcare14751475GLM 5.2
LMArena Text: legal and government14861473GLM 5.2
LMArena Text: business and finance14511460GLM 5.3 Flash
Neither model has a result on the other 7 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index35
Web design benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
Design Arena: websites12851272GLM 5.2
Design Arena: UI components12961316GLM 5.3 Flash
Design Arena: game dev12681279GLM 5.3 Flash
Design Arena: data viz12851262GLM 5.2
Design Arena: 3D12881316GLM 5.3 Flash
Design Arena: all code categories12881282GLM 5.2
Design Arena: SVG12141284GLM 5.3 Flash
Design Arena: ASCII art11901269GLM 5.3 Flash
LiveBench designNot measured77.2%Only GLM 5.3 Flash measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index119
General benchmarks for GLM 5.2 and GLM 5.3 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
Vals IndexNot measured50.9%Only GLM 5.3 Flash measured
LiveBench data analysisNot measured76.4%Only GLM 5.3 Flash measured
MedScribe (Vals AI)83.5%88.9%GLM 5.3 Flash
IFBench73.3%Not measuredOnly GLM 5.2 measured
Long context reasoning (AA-LCR)78.3%80.0%GLM 5.3 Flash
DTBench93.6%Not measuredOnly GLM 5.2 measured
LMCA45.8%Not measuredOnly GLM 5.2 measured
Surface Evolver Bench55.6%52.5%GLM 5.2
LMArena Text14701471GLM 5.3 Flash
LMArena Text: instruction following14661478GLM 5.3 Flash
LMArena Text: creative writing14631441GLM 5.2
LMArena VisionNot measured1294Only GLM 5.3 Flash measured
LMArena Text: English14751474GLM 5.2
LMArena Text: non-English14591462GLM 5.3 Flash
LMArena Text: Chinese15221531GLM 5.3 Flash
LMArena Text: German14701462GLM 5.2
LMArena Text: Japanese14501444GLM 5.2
LMArena Text: Korean14401436GLM 5.2
LMArena Text: French14791491GLM 5.3 Flash
LMArena Text: Spanish14781475GLM 5.2
LMArena Text: Russian14681465GLM 5.2
LMArena Text: Polish14561450GLM 5.2
LMArena Text: multi-turn14691469GLM 5.3 Flash
LMArena Text: longer queries14801481GLM 5.3 Flash
LMArena Text: writing and language14631455GLM 5.2
LMArena Text: entertainment and media14501440GLM 5.2
LMArena Vision: EnglishNot measured1295Only GLM 5.3 Flash measured
LMArena Vision: ChineseNot measured1363Only GLM 5.3 Flash measured
LMArena Vision: reading textNot measured1299Only GLM 5.3 Flash measured
LMArena Vision: humorNot measured1279Only GLM 5.3 Flash measured
LMArena Vision: creative writingNot measured1270Only GLM 5.3 Flash measured
Neither model has a result on the other 15 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 Flash
BenchmarkGLM 5.2GLM 5.3 FlashLeader
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 Flash$0.0004Fits in one request

GLM 5.3 Flash costs less

Repository review

50K input + 3K output tokens

GLM 5.2$0.023Fits in one request
GLM 5.3 Flash$0.0090Fits in one request

GLM 5.3 Flash costs less

Agent loop

200K input + 10K output tokens

GLM 5.2$0.077Fits in one request
GLM 5.3 Flash$0.035Fits in one request

GLM 5.3 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 GLM 5.2 and GLM 5.3 Flash
SpecificationZ.ai GLM 5.2Z.ai GLM 5.3 Flash
MakerZ.aiZ.ai
WeightsOpen weightsOpen weights
ReleasedJune 16, 2026August 26, 2026
Context window1.05M1.05M
Input $/Mtok$0.0341$0.15
Output $/Mtok$7.00$0.50
Coding index57.771.0
Independent measurements behind it12
Cost per solved task$0.86$0.02
SourceOpenRouter listingOpenRouter listing

Questions

Is GLM 5.2 better than GLM 5.3 Flash for coding?

Pick GLM 5.3 Flash for coding: it scores 71.0 on the Vibe Coding Index against 57.7 for GLM 5.2. Pick GLM 5.2 when what matters is Vibe Code Bench (Vals AI) or Code migration (Vals AI), where it leads.

Which is cheaper, GLM 5.2 or GLM 5.3 Flash?

GLM 5.2 costs $0.0341 per million input tokens and $7.00 per million output tokens; GLM 5.3 Flash costs $0.15 and $0.50. Per solved SWE-bench Verified task, GLM 5.2 costs $0.86 and GLM 5.3 Flash $0.02.

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

Neither: both take 1.05M tokens.

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

76 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 Flash's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next