BenchLeader

GLM 5.2 vs GLM 5.3

Full profiles: GLM 5.2 (Zhipu AI) and GLM 5.3 (Zhipu AI): every benchmark, price and rank.

Updated · Data from 11 sources

As of 16 Nov 2026, GLM 5.2 and GLM 5.3 are neck and neck on quality (58.9 vs 58.8 on the BenchLeader Index, inside the ±5.0 margin), and they cost about the same at $2.15 vs $2.15 per 1M tokens blended. Choose 5.2 for agents & tools (62 vs 58), reasoning (57 vs 55), and coding (62 vs 60) and fast first answers (26.8 s vs 57.6 s); choose 5.3 for knowledge (54 vs 44) and maths (60 vs 53).

Key facts (as of 16 Nov 2026)

  • GLM 5.2 scores 58.9 on the BenchLeader Index and GLM 5.3 scores 58.8 (BenchLeader, 16 Nov 2026).
  • GLM 5.2 costs $2.15 per 1M tokens blended (list price, 16 Nov 2026).
  • GLM 5.3 costs $2.15 per 1M tokens blended (list price, 16 Nov 2026).
  • Of the 51 benchmarks both models have, GLM 5.3 leads 45 (BenchLeader, 16 Nov 2026).
  • On Arena WebDev, GLM 5.2 is rated 1608 and GLM 5.3 1623 (LMArena, as published by 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingGLM 5.262 vs 60 category score; Arena WebDev 1608 vs 1623Medium
Agents & tool useGLM 5.262 vs 58 category score; τ³-Banking (AA) 34.6% vs 50.3%Medium
MathsGLM 5.360 vs 53 category score; FrontierMath Tiers 1–3 68.8% vs 59.2%Medium
ReasoningGLM 5.257 vs 55 category score; GPQA Diamond 91.9% vs 90.9%Medium
KnowledgeGLM 5.354 vs 44 category score; SimpleQA Verified 41.0% vs 34.2%High
Long documentsEither64 vs 65 category score; AA-LCR v1.1 78.3% vs 79.7%Either
Chat quality (human preference)Either66 vs 67 on Arena votesEither
Lowest costEither$2.15 vs $2.15 per 1M blended; agentic coding session $0.102 vs $0.102–
Fastest first answerGLM 5.226.8 s (GLM 5.2 (max)) vs 57.6 s (GLM 5.3 (max)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • GLM 5.2 (max) and GLM 5.3 (max) are neck and neck on quality (58.9 vs 58.8, within the ±5.0 margin).
  • GLM 5.2 (max) is stronger in agents & tools, coding, reasoning.
  • GLM 5.3 (max) is stronger in human preference, knowledge, long context, maths.
  • They cost about the same ($2.15 per 1M blended).

At a glance

SpecGLM 5.2GLM 5.3
DeveloperZhipu AIZhipu AI
Released16 Jun 202618 Aug 2026
Context window (tokens)1,000,0001,000,000
Compared atGLM 5.2 (max)GLM 5.3 (max)
Open weightsYesYes
Price in / out / cached input per 1M$1.40 / $4.40 / $0.260$1.40 / $4.40 / $0.260

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryGLM 5.2 (max)GLM 5.3 (max)Leader
Reasoning57.254.7GLM 5.2
Coding61.859.7GLM 5.2
Agents & tools62.157.6GLM 5.2
Maths52.660.0GLM 5.3
Knowledge44.353.5GLM 5.3
Instruction following71.8–not comparable
Human preference66.467.2level
Long context63.864.5level

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsKnowledgeLong contextHumanpreferenceGLM 5.2, Reasoning: 57.2GLM 5.2, Coding: 61.8GLM 5.2, Agents & tools: 62.1GLM 5.2, Maths: 52.6GLM 5.2, Knowledge: 44.3GLM 5.2, Long context: 63.8GLM 5.2, Human preference: 66.4GLM 5.3, Reasoning: 54.7GLM 5.3, Coding: 59.7GLM 5.3, Agents & tools: 57.6GLM 5.3, Maths: 60.0GLM 5.3, Knowledge: 53.5GLM 5.3, Long context: 64.5GLM 5.3, Human preference: 67.2

Not drawn, measured for fewer than two of these models: Instruction following, Multimodal.

Model, click to add or removeIndex

Where sources disagree

Across the 51 benchmarks both have, GLM 5.2 leads 5, GLM 5.3 leads 45 and 1 is a tie. Every overall index that rates both puts GLM 5.3 ahead.

Overall indexGLM 5.2 (max)GLM 5.3 (max)Leader
BenchLeader Index58.958.8neck and neck
Arena Text14751482GLM 5.3

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureGLM 5.2 (max)GLM 5.3 (max)
Blended price per 1M tokens$2.15$2.15
Input / output per 1M$1.40 / $4.40$1.40 / $4.40
Cached input per 1M$0.260$0.260
Price per task: chat reply$0.0019$0.0019
Price per task: summarize a 30-page report$0.019$0.019
Price per task: code edit$0.015$0.015
Price per task: agentic coding session$0.102$0.102
Price per task: structured extraction$0.0037$0.0037
Agentic coding session, input mostly cached$0.050$0.050
Cost per benchmark task (AA Index)–$5.01
Output tokens per task (AA Index)–119k
Of which reasoning–85k
Time per task (AA Index)–23 min
Output speed69 tok/s88 tok/s
Time to first answer26.8 s57.6 s
Time to a complete answer, last measured 13 Nov 202656.7 s144.1 s

Each model is compared at its best-ranked reasoning-effort setting. GLM 5.2 (max) scores 58.9 and takes 27 s to its first answer token; GLM 5.2 scores 51.9 in 1 s, GLM 5.2 (no reasoning) scores 45.8 in 2 s. GLM 5.3 (max) scores 58.8 and takes 58 s to its first answer token; GLM 5.3 (low) scores 47.3 in 7 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

162 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherGLM 5.2 (max)GLM 5.3 (max)
BenchLeader IndexBenchLeader58.958.8
Agents & tools scoreBenchLeader62.157.6
Coding scoreBenchLeader61.859.7
Human preference scoreBenchLeader66.467.2
Instruction following scoreBenchLeader71.8–
Knowledge scoreBenchLeader44.353.5
Long context scoreBenchLeader63.864.5
Maths scoreBenchLeader52.660.0
Reasoning scoreBenchLeader57.254.7
Blended price $/MBenchLeader$2.15$2.15
Output speedBenchLeader69 tok/s88 tok/s
Time to first answerBenchLeader26.8 s57.6 s
Context windowBenchLeader1,000,000 tokens1,000,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub91.9%90.9%
SimpleBenchSimpleBench58.8%†66.2%†
Arena Hard PromptsArena (LMArena)14961507
LiveBench 1.0 ReasoningLiveBench78.6%†85.8%†
GPQA Diamond (AA)Artificial Analysis89.5%91.7%
Humanity's Last Exam (AA)Artificial Analysis41.1%42.3%
HLE-Diamond (AA)Artificial Analysis17.7%17.9%
GPQA Diamond (Vals)Vals AI85.6%†88.1%
Kagi LLM BenchmarkKagi LLM Benchmark62.6%†–
ARC-AGI-1ARC Prize77.0%†–
ARC-AGI-2ARC Prize22.8%†–
CritPtArtificial Analysis20.9%19.1%
CritPt v2 hard subset (AA)Artificial Analysis–33.5%
MysteryMechanismVals AI–23.0%
HLE-Diamond (Scale)Scale AI SEAL–16.4%†
Arena ExpertArena (LMArena)15021522
Chess PuzzlesEpoch AI Benchmarking Hub22.0%37.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub18.0%†33.0%
LMCAEpoch AI Benchmarking Hub45.8%55.5%†
DTBenchEpoch AI Benchmarking Hub93.6%87.7%†
BTF-3Epoch AI Benchmarking Hub–14.9%†
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub78.7%–
SciCodeSciCode51.2%59.0%
WeirdMLWeirdML70.1%75.4%
FrontierCodeCognition24.5%†40.1%
Show all 162 benchmarks
MetricPublisherGLM 5.2 (max)GLM 5.3 (max)
Coding
Arena CodingArena (LMArena)15151522
Arena WebDevArena (LMArena)16081623
LiveBench 1.0 CodingLiveBench79.7%†79.0%†
LiveCodeBenchVals AI69.5%†80.5%
IOIVals AI–68.4%
SWE-bench (Vals)Vals AI82.8%95.4%
Code MigrationVals AI37.9%44.2%
Code Migration v1.1Vals AI–55.1%
ProgramBenchVals AI0.5%1.5%
Vibe Code Bench 1-100Vals AI–20.0%
Vibe Code Bench v1.1Vals AI64.0%78.1%
Vibe Code Bench 1-100 v1.1Vals AI–29.8%
SWE Atlas v1.0: Codebase QnAScale AI SEAL48.1%†–
SWE Atlas v1.0: RefactoringScale AI SEAL42.4%†–
SWE Atlas v1.0: Test WritingScale AI SEAL41.5%†–
SWE Atlas v1.1Scale AI SEAL–58.1%†
SWE Atlas v1.1: Codebase QnAScale AI SEAL–70.2%†
SWE Atlas v1.1: Test WritingScale AI SEAL–55.6%†
SWE Atlas v1.1: RefactoringScale AI SEAL–48.6%†
SWE-Bench Pro V2Scale AI SEAL–95.6%
SWE-Bench Pro V2 HARDScale AI SEAL–84.3%
MirrorCodeEpoch AI Benchmarking Hub36.7%–
Surface Evolver BenchEpoch AI Benchmarking Hub55.6%†–
DeepSWE v1.1Epoch AI Benchmarking Hub43.8%69.0%
CursorBenchCursor–42.6%
ALE-BenchEpoch AI Benchmarking Hub1010.21317.4†
FrontierSWEEpoch AI Benchmarking Hub–30.2%
Terminal-Bench 4.0 (AA)Artificial Analysis1.0%41.9%
Terminal-Bench 2.1 (AA)Artificial Analysis77.9%83.9%
Startrise FrontendStartrise72.1%†73.2%†
GameBench 2 OverallDuelLab GameBench 243.1†64.6†
GameBench 2 by reasoning settingDuelLab GameBench 232.4†72.3
Design Arena WebsiteDesign Arena1277†1279†
FrontierCode 1.1 (AA)Artificial Analysis–40.1%
FrontierCode 2.0 (AA)Artificial Analysis–21.6%
Agents & tools
Terminal-BenchTerminal-Bench–41.2%
APEX-AgentsMercor45.2%†56.6%†
Arena AgentArena (LMArena)0.21
LiveBench Agentic CodingLiveBench51.8%†60.9%†
Terminal-Bench HardArtificial Analysis50.8%–
τ²-Bench Telecom (AA)Artificial Analysis99.1%–
Terminal-Bench 2.1 (Vals)Vals AI67.8%71.5%
MCP AtlasScale AI SEAL77.8%†84.2%†
HiL-BenchScale AI SEAL43.7%†58.3%†
τ³-bench Bankingτ²-bench37.1%†–
GDPval-AA v2.1, normalised scoreArtificial Analysis–57.5%
GDPval-AA v2.1Artificial Analysis13801655
τ³-Banking (AA)Artificial Analysis34.6%50.3%
ITBench SRE (AA)Artificial Analysis42.7%46.1%
Analyst Agent (AA)Artificial Analysis–38.8%
APEX-Agents (AA)Artificial Analysis33.7%–
CyberBenchVals AI–72.1%
Finance Agent v2Vals AI49.7%†55.8%
Finance Agent v2.1Vals AI–65.0%
Harvey's Legal Agent Benchmark v1.1Vals AI–14.6%
Harvey's Legal Agent Benchmark v1Vals AI7.1%8.3%
Legal Research BenchVals AI31.3%49.0%
SkillsBenchVals AI45.1%47.5%
SREBenchVals AI0.0%9.2%
Tax Agent BenchVals AI–43.3%
Terminal-Bench 4.0 (Vals)Vals AI–38.9%
Terminal-Bench ScienceVals AI–5.7%
Arena Agent Task CompletionArena (LMArena)1.73.1
PostTrainBenchEpoch AI Benchmarking Hub30.4%34.5%
GBAEvalEpoch AI Benchmarking Hub0.0%†–
Vending-Bench 2Andon Labs8313.8†8163.6†
AutomationBenchArtificial Analysis28.4%62.2%
GDP.pdf (AA)Artificial Analysis10.4%11.2%
Harvey LAB-AA v1.1Harvey–0.3%
EnterpriseOps-GymArtificial Analysis42.7%36.4%
AA-Briefcase v1.1Artificial Analysis12261501
GDPxGDPx by SemiAnalysis–2.7%
Design Arena AgenticDesign Arena1194†1239†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis–9.5%
CWE-Bench (AA)Artificial Analysis–55.8%
DeepSec-Bench (AA)Artificial Analysis–24.3%
CyberGym end-to-end (AA)Artificial Analysis–29.0%
FACTS SearchGoogle–87.7%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub59.2%68.8%
FrontierMath Tier 4Epoch AI Benchmarking Hub29.3%29.3%
OTIS Mock AIMEEpoch AI Benchmarking Hub86.4%91.1%
ProofBenchVals AI35.0%49.0%
LiveBench MathematicsLiveBench89.8%†87.9%†
AIME 2026MathArena90.0%–
HMMT February 2026MathArena92.4%–
MathArena ApexMathArena19.8%–
Arena MathsArena (LMArena)14871501
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub34.2%41.0%
LiveBench 1.0 Data AnalysisLiveBench73.7%†70.2%†
AA-Omniscience v1.1Artificial Analysis0.66.0
MMLU-ProVals AI86.7%†86.8%
LegalBenchVals AI84.1%†84.8%
CorpFinVals AI66.1%†–
TaxEvalVals AI73.3%†72.4%
Excel Modeling BenchmarkVals AI61.5%†56.3%
Excel Modeling Benchmark v1.1Vals AI–69.1%
MedCodeVals AI40.8%†42.9%
MedScribeVals AI83.5%†88.8%
Public Benefits Bench v1.1Vals AI–68.5%
Access to Justice BenchmarkVals AI–73.7%
Financial Coaching BenchVals AI–74.1%
Immigration BenchVals AI–66.7%
DrugDiscoveryBenchScale AI SEAL36.2%†–
EBR-benchEpoch AI Benchmarking Hub9.5%–
AA-Omniscience v1.1: accuracyArtificial Analysis21.1%32.3%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis74.1%61.1%
HealthBench Professional (AA)OpenAI–52.8%
FACTS ParametricGoogle–40.3%†
Instruction following
LiveBench LanguageLiveBench76.2%†79.9%†
LiveBench Instruction FollowingLiveBench62.3%†69.3%†
IFBenchArtificial Analysis73.3%–
Arena Instruction FollowingArena (LMArena)14701481
Human preference
Arena TextArena (LMArena)14751482
EQ-Bench 4EQ-Bench1208†–
EQ-Bench Creative Writing v3EQ-Bench1759†2076†
Arena Creative WritingArena (LMArena)14511456
Arena Multi-turnArena (LMArena)14771482
Arena Longer QueriesArena (LMArena)14861488
Multimodal
Code2Video-AAArtificial Analysis–1034
AA-3DArtificial Analysis–794
Long context
AA-LCR v1.1Artificial Analysis78.3%79.7%
MLCRArtificial Analysis12.8%17.2%
FACTS GroundingGoogle–55.2%†
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub153.0†157.1†
LiveBench 1.0LiveBench73.2%†76.1%†
AA Intelligence Index v5.0.1Artificial Analysis–31.6
AA Intelligence Index v4.3.2Artificial Analysis33.744.8
AA Progress Index v1Artificial Analysis302.5322.7
Vals IndexVals AI–59.9
Vals Finance IndexVals AI–61.4%
Vals RSI IndexVals AI–22.9%
Vals Router IndexVals AI–59.7%†
Vals Legal IndexVals AI–74.6%
AA Cyber IndexArtificial Analysis–36.4
Safety & honesty
FORTRESSScale AI SEAL48.7%†28.2%
DistressBenchScale AI SEAL–72.5%†
Metadata
AA Openness IndexArtificial Analysis44.433.3

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • GLM 5.2 (max): 1 new result (GDPval-AA v2.1).
  • GLM 5.2 (max): BenchLeader Index 59.5 → 59.0.
  • GLM 5.2 (max): 3 new results (AIME 2026, HMMT February 2026, MathArena Apex); 1 score revision.
  • GLM 5.3 (max): 1 new result (PostTrainBench).
  • GLM 5.2 (max): 1 new result (MirrorCode).
  • GLM 5.3 (max): 4 score revisions.
  • GLM 5.2 (max): BenchLeader Index 60.2 → 59.5.
  • GLM 5.2 (max): 4 new results (HLE-Diamond (AA), AA-Omniscience v1.1, AA-Omniscience v1.1: accuracy…).
  • GLM 5.2 (max): cached input price moved from $0.30/M to $0.26/M (13% down).
  • GLM 5.3 (max): 1 score revision; cached input price moved from $0.30/M to $0.26/M (13% down).

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

GLM 5.2 vs GLM 5.3: questions

Is GLM 5.2 better than GLM 5.3?
GLM 5.2 (max) and GLM 5.3 (max) are neck and neck on quality (58.9 vs 58.8, within the ±5.0 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
GLM 5.2 vs GLM 5.3: which is cheaper, and will that change?
They cost the same: $2.15 per million tokens blended. No price change has been announced for either.
Is GLM 5.2 or GLM 5.3 better for coding?
GLM 5.2 scores higher in coding (62 vs 60 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1608 for GLM 5.2 against 1623 for GLM 5.3.
Which is faster, GLM 5.2 or GLM 5.3?
GLM 5.3 streams faster (88 vs 69 output tokens per second, Artificial Analysis 7-day median).
Is GLM 5.2 or GLM 5.3 better for agents and tool use?
GLM 5.2 scores higher in agentic tasks (62 vs 58 on the category index, where 50 is average). On τ³-Banking (AA), the top shared agents & tools benchmark, it is 34.6% for GLM 5.2 against 50.3% for GLM 5.3.
More questions (4)
Which is better for long documents?
They are level on long-context benchmarks (64 vs 65). Both accept 1,000,000 tokens of context.
Which has lower latency?
GLM 5.2 answers first: 26.8 s against 57.6 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on GLM 5.2 vs GLM 5.3?
Across the 51 benchmarks both have, GLM 5.2 leads 5, GLM 5.3 leads 45 and 1 is a tie. Every overall index that rates both puts GLM 5.3 ahead.
When was GLM 5.3 released?
GLM 5.3 was released on 18 Aug 2026.