LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

GLM 5.2 vs Grok 4.7

Updated October 21, 2026. 68 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 Grok 4.7
GLM 5.2Z.ai
Coding
57.7
Overall
33.7
Output $/M
$7.00
36
Grok 4.7xAI
Coding
75.0
Overall
46.4
Output $/M
$6.00
31
vs boards won of 68 shared, 1 tied

Decision reading

Pick Grok 4.7 for coding: it scores 75.0 on the Vibe Coding Index against 57.7 for GLM 5.2.

Pick GLM 5.2 when the context window is larger, 1.05M against 500K tokens, or when what matters is LMArena Text: coding or LMArena Text: software and IT, where it leads.

Model AGLM 5.2Z.ai57.7/1001 independent benchmark$0.86 per solved taskModel BGrok 4.7xAI75.0/1002 independent benchmarksNo cost per solved task
Shared results
68
GLM 5.2 only
29
Grok 4.7 only
31
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, Grok 4.7 75.0 from 2 benchmarks.
Widest category
Reasoning, 12.4 points apart on average over 9 shared benchmarks.
Widest single gap
ARC-AGI-2: GLM 5.2 22.8% against Grok 4.7 61.4%, 38.6 points apart on a 0 to 100 scale.
Output price
GLM 5.2 $7.00 against Grok 4.7 $6.00 per million output tokens, 1.2x apart.
Context window
GLM 5.2 takes 1.05M tokens, Grok 4.7 500K.

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 5.2BarsGrok 4.7Ties
Coding281 tie
Intelligence02
Agentic work06
Reasoning36
Math51
Knowledge43
Web design51
General174
All shared boards36311 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index28
Coding benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
LiveBench codingNot measured84.5%Only Grok 4.7 measured
LMArena WebDev16041637Grok 4.7
AA Coding Index68.8Not measuredOnly GLM 5.2 measured
LiveCodeBench (Vals AI)69.5%Not measuredOnly GLM 5.2 measured
SWE-bench Verified (Vals AI)82.8%Not measuredOnly GLM 5.2 measured
Terminal-Bench 4 (Vals AI)Not measured28.8%Only Grok 4.7 measured
Vibe Code Bench (Vals AI)64.0%86.2%Grok 4.7
Code migration (Vals AI)37.9%44.8%Grok 4.7
ProgramBench (Vals AI)0.5%0.5%Tie
IOI (Vals AI)Not measured57.7%Only Grok 4.7 measured
Terminal-Bench 2.1 (Vals AI)67.8%73.4%Grok 4.7
SWE-bench Multimodal (Vals AI)Not measured42.9%Only Grok 4.7 measured
SkillsBench (Vals AI)45.1%Not measuredOnly GLM 5.2 measured
CyberBench (Vals AI)Not measured69.5%Only Grok 4.7 measured
SciCode51.2%57.8%Grok 4.7
Terminal-Bench Hard (Artificial Analysis run)50.8%Not measuredOnly GLM 5.2 measured
FrontierCode24.5%47.6%Grok 4.7
FrontierSWENot measured29.5%Only Grok 4.7 measured
WeirdML70.1%Not measuredOnly GLM 5.2 measured
ALE-Bench1047.0Not measuredOnly GLM 5.2 measured
LMArena Text: coding14871432GLM 5.2
LMArena Text: software and IT14861421GLM 5.2
LMArena WebDev: React16171663Grok 4.7
LMArena WebDev: HTML15651627Grok 4.7
LMArena WebDev: from an imageNot measured1586Only Grok 4.7 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 Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
AA Intelligence Index33.746.4Grok 4.7
Epoch Capabilities Index151.9154.4Grok 4.7
Agentic workLike-for-likePublic boards, shown as published, kept out of the index06
Agentic work benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
AA Agentic Index38.4Not measuredOnly GLM 5.2 measured
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 measured86.4%Only Grok 4.7 measured
LiveBench browserNot measured60.0%Only Grok 4.7 measured
LiveBench long agenticNot measured50.1%Only Grok 4.7 measured
SREBench (Vals AI)0.0%3.8%Grok 4.7
Finance Agent (Vals AI)49.7%52.3%Grok 4.7
Tax Agent Bench (Vals AI)Not measured34.2%Only Grok 4.7 measured
Legal Research Bench (Vals AI)31.3%47.1%Grok 4.7
Harvey's Legal Agent Benchmark (Vals AI)Not measured18.3%Only Grok 4.7 measured
Excel Modeling Benchmark (Vals AI)61.5%67.0%Grok 4.7
BioMysteryBench (Vals AI)Not measured69.3%Only Grok 4.7 measured
Terminal-Bench-Science (Vals AI)Not measured10.0%Only Grok 4.7 measured
Vals Science IndexNot measured34.8%Only Grok 4.7 measured
Robo-bench (Vals AI)Not measured0.2%Only Grok 4.7 measured
τ²-Bench Telecom99.1%Not measuredOnly GLM 5.2 measured
APEX-Agents45.2%54.6%Grok 4.7
PostTrainBench31.7%Not measuredOnly GLM 5.2 measured
Vending-Bench 2$8,314$10,537Grok 4.7
Neither model has a result on the other 13 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index36
Reasoning benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
GPQA Diamond85.6%Not measuredOnly GLM 5.2 measured
LiveBench reasoningNot measured82.7%Only Grok 4.7 measured
MysteryMechanism (Vals AI)Not measured25.2%Only Grok 4.7 measured
GPQA Diamond (Artificial Analysis run)89.5%Not measuredOnly GLM 5.2 measured
Humanity's Last Exam (Artificial Analysis run)41.1%43.1%Grok 4.7
GPQA Diamond (Epoch AI run)91.9%92.7%Grok 4.7
ARC-AGI77.0%90.2%Grok 4.7
ARC-AGI-222.8%61.4%Grok 4.7
SimpleBench58.8%Not measuredOnly GLM 5.2 measured
Chess puzzles22.0%41.0%Grok 4.7
Mystery game puzzles19.0%29.0%Grok 4.7
EBR-bench9.5%Not measuredOnly GLM 5.2 measured
LMArena Text: hard prompts14811415GLM 5.2
LMArena Text: expert14871439GLM 5.2
LMArena Text: hard prompts in English14811427GLM 5.2
LMArena Vision: diagramsNot measured1254Only Grok 4.7 measured
LMArena Vision: homeworkNot measured1309Only Grok 4.7 measured
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index51
Math benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
OTIS Mock AIME86.4%98.1%Grok 4.7
FrontierMath, tiers 1 to 359.2%53.0%GLM 5.2
FrontierMath, tier 429.3%17.1%GLM 5.2
ProofBench35.0%26.0%GLM 5.2
LMArena Text: math14831414GLM 5.2
LMArena Text: mathematical work14891435GLM 5.2
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index43
Knowledge benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
MMLU Pro86.7%Not measuredOnly GLM 5.2 measured
LegalBench (Vals AI)84.1%84.4%Grok 4.7
CorpFin (Vals AI)66.1%Not measuredOnly GLM 5.2 measured
TaxEval (Vals AI)73.3%Not measuredOnly GLM 5.2 measured
MedCode (Vals AI)40.8%49.5%Grok 4.7
SimpleQA Verified34.2%56.0%Grok 4.7
LMArena Text: science14891419GLM 5.2
LMArena Text: medicine and healthcare14751395GLM 5.2
LMArena Text: legal and government14861423GLM 5.2
LMArena Text: business and finance14511389GLM 5.2
Neither model has a result on the other 7 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index51
Web design benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
Design Arena: websites12851255GLM 5.2
Design Arena: UI components12961271GLM 5.2
Design Arena: game dev12681314Grok 4.7
Design Arena: data viz12851227GLM 5.2
Design Arena: 3D12881269GLM 5.2
Design Arena: all code categories12881271GLM 5.2
Design Arena: SVG1214Not measuredOnly GLM 5.2 measured
Design Arena: ASCII art1190Not measuredOnly GLM 5.2 measured
LiveBench designNot measured78.0%Only Grok 4.7 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index174
General benchmarks for GLM 5.2 and Grok 4.7
BenchmarkGLM 5.2Grok 4.7Leader
Vals IndexNot measured58.6%Only Grok 4.7 measured
LiveBench data analysisNot measured76.9%Only Grok 4.7 measured
Public Benefits Bench (Vals AI)Not measured65.6%Only Grok 4.7 measured
SAGE (Vals AI)Not measured40.8%Only Grok 4.7 measured
MedScribe (Vals AI)83.5%89.4%Grok 4.7
IFBench73.3%Not measuredOnly GLM 5.2 measured
Long context reasoning (AA-LCR)78.3%80.0%Grok 4.7
DTBench93.6%96.0%Grok 4.7
LMCA45.8%49.4%Grok 4.7
Surface Evolver Bench55.6%Not measuredOnly GLM 5.2 measured
Furniture assemblyNot measured20.8%Only Grok 4.7 measured
LMArena Text14701399GLM 5.2
LMArena Text: instruction following14661406GLM 5.2
LMArena Text: creative writing14631381GLM 5.2
LMArena VisionNot measured1231Only Grok 4.7 measured
LMArena Text: English14751411GLM 5.2
LMArena Text: non-English14591386GLM 5.2
LMArena Text: Chinese15221449GLM 5.2
LMArena Text: German14701366GLM 5.2
LMArena Text: Japanese14501359GLM 5.2
LMArena Text: Korean14401383GLM 5.2
LMArena Text: French14791431GLM 5.2
LMArena Text: Spanish14781406GLM 5.2
LMArena Text: Russian14681399GLM 5.2
LMArena Text: Polish14561420GLM 5.2
LMArena Text: multi-turn14691394GLM 5.2
LMArena Text: longer queries14801416GLM 5.2
LMArena Text: writing and language14631394GLM 5.2
LMArena Text: entertainment and media14501376GLM 5.2
LMArena Vision: EnglishNot measured1229Only Grok 4.7 measured
LMArena Vision: ChineseNot measured1250Only Grok 4.7 measured
LMArena Vision: reading textNot measured1241Only Grok 4.7 measured
LMArena Vision: creative writingNot measured1217Only Grok 4.7 measured
Neither model has a result on the other 13 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 Grok 4.7
BenchmarkGLM 5.2Grok 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

GLM 5.2$0.0035Fits in one request
Grok 4.7$0.0050Fits in one request

GLM 5.2 costs less

Repository review

50K input + 3K output tokens

GLM 5.2$0.023Fits in one request
Grok 4.7$0.118Fits in one request

GLM 5.2 costs less

Agent loop

200K input + 10K output tokens

GLM 5.2$0.077Fits in one request
Grok 4.7$0.460Fits in one request

GLM 5.2 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 GLM 5.2 and Grok 4.7
SpecificationZ.ai GLM 5.2SpaceXAI Grok 4.7
MakerZ.aixAI
WeightsOpen weightsNo public weights listed
ReleasedJune 16, 2026September 21, 2026
Context window1.05M500K
Input $/Mtok$0.0341$2.00
Output $/Mtok$7.00$6.00
Coding index57.775.0
Independent measurements behind it12
Cost per solved task$0.86Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GLM 5.2 better than Grok 4.7 for coding?

Pick Grok 4.7 for coding: it scores 75.0 on the Vibe Coding Index against 57.7 for GLM 5.2. Pick GLM 5.2 when the context window is larger, 1.05M against 500K tokens, or when what matters is LMArena Text: coding or LMArena Text: software and IT, where it leads.

Which is cheaper, GLM 5.2 or Grok 4.7?

GLM 5.2 costs $0.0341 per million input tokens and $7.00 per million output tokens; Grok 4.7 costs $2.00 and $6.00. Cost per solved task is known for GLM 5.2 only.

Which has the bigger context window, GLM 5.2 or Grok 4.7?

GLM 5.2, with 1.05M tokens against 500K.

How many benchmarks is the GLM 5.2 vs Grok 4.7 comparison based on?

68 benchmarks have a result for both models as of October 21, 2026. GLM 5.2's coding index rests on 1 independent benchmark and Grok 4.7's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next