Inference360
Z.AI

GLM-5.2

GLM-5.2 is a reasoning model from Z.AI in the GLM-5 family. 48 benchmarks count toward its score, in 6 categories.

availableShows if the model has enough results for an index.
IndexOverall score. 50 is the middle.62.7 ±3.8
CoverageShare of the index weight with results.85%
SpeedOutput tokens per second.74/s
Input / 1MUS dollars per 1M input tokens.$0.65
Output / 1MUS dollars per 1M output tokens.$2.04
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1467 (#29)

50 is the middle of the board. The range shows the doubt in the index. A free tier is available.

36,798 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category. 50 is the middle.

50 is the middle
AgenticMulti-step tasks with tools.
64.3
CodingCode writing and repair.
64.2
ReasoningLogic problems and puzzles.
59.7
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
60.3
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
61.0
MathMath problems.
61.9

Results

48 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result on the index scale.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
AIME 2026Math99.2%57.3——Qwen
τ²-Bench Tool-Agent-User EvaluationAgentic99.1%69.9——Victor Barres et al.
Harvard-MIT Mathematics Tournament November 2025Math94.4%———Qwen
Harvard-MIT Mathematics Tournament February 2026Math92.5%59.4——Qwen
GPQA diamondKnowledge91.9%63.0max effort—Epoch AI
Graduate-Level Google-Proof Q&AKnowledge91.2%62.4——David Rein et al.
GPQA DiamondKnowledge91.2%62.4——David Rein et al.
MMAnswerBenchMath91.0%———Qwen
LiveBench MathematicsMath89.8%67.9—25 Jun 2026LiveBench
Artificial Analysis GPQA DiamondKnowledge89.5%60.3——Artificial Analysis
MMLU ProKnowledge86.7%57.2—1 Sept 2026Vals AI
OTIS Mock AIME 2024-2025Math86.4%59.8max effort—Epoch AI
GPQA DiamondKnowledge85.6%57.3—1 Sept 2026Vals AI
SWE-benchCoding82.8%64.0—1 Sept 2026Vals AI
LiveBench CodingCoding79.7%69.9—25 Jun 2026LiveBench
SWE-Bench verifiedCoding78.7%60.7max effort—Epoch AI
LiveBench ReasoningReasoning78.6%65.2—25 Jun 2026LiveBench
Artificial Analysis Long Context ReasoningReasoning78.3%62.4——Artificial Analysis
ARC-AGI-1 (semi-private)Reasoning77.0%63.6——ARC Prize Foundation
MCP AtlasAgentic76.8%66.0——OpenAI
LiveBench LanguageKnowledge76.2%62.6—25 Jun 2026LiveBench
LiveBench Data AnalysisReasoning73.7%58.4—25 Jun 2026LiveBench
Artificial Analysis IFBenchInstruction73.3%63.5——Artificial Analysis
LiveCodeBenchCoding69.5%47.5—1 Sept 2026Vals AI
Artificial Analysis Coding IndexCoding68.8%67.4——Artificial Analysis
Terminal-Bench 2.1Agentic67.8%64.1—21 Sept 2026Vals AI
Vibe Code Bench v1.1Coding64.0%68.8OpenHands21 Sept 2026Vals AI
ProgramBench: Can Language Models Rebuild Programs From Scratch?Coding63.7%65.5——John Yang et al.
LiveBench Instruction FollowingInstruction62.3%58.5—25 Jun 2026LiveBench
SWE-bench ProCoding62.1%64.1——Xiang Deng et al.
FrontierMath-Tiers-1-3-v2-PrivateMath59.2%64.8max effort—Epoch AI
OpenHarmony Bench v1.0Coding58.4%67.5——OpenHarmony Bench authors
cursorBench32Coding55.0%63.8——Benchmark authors
Humanity's Last ExamKnowledge54.7%75.1——Center for AI Safety et al.
LiveBench Agentic CodingAgentic51.8%65.4—25 Jun 2026LiveBench
Artificial Analysis SciCodeCoding51.2%63.9——Artificial Analysis
NL2RepoCoding48.9%63.4——MiniMax
ToolathlonAgentic48.2%62.9——OpenAI
SkillsBenchCoding45.1%61.2OpenHands11 Sept 2026Vals AI
GDPval-AA normalizedAgentic42.9%68.5——Artificial Analysis
Artificial Analysis ITBench-AAAgentic42.7%———Artificial Analysis
Artificial Analysis Humanity's Last ExamKnowledge41.1%69.1——Artificial Analysis
Humanity's Last Exam without toolsKnowledge40.5%63.1——OpenAI
Artificial Analysis Agentic IndexAgentic39.4%69.3——Artificial Analysis
Code MigrationCoding37.9%69.2—21 Sept 2026Vals AI
τ²-bench BankingAgentic37.1%25.5xhigh effort · Sierra4 Aug 2026Sierra Research
SimpleQA VerifiedKnowledge34.2%53.2max effort—Epoch AI
Artificial Analysis Intelligence IndexKnowledge33.7%64.6——Artificial Analysis
APEX-Agents-AAAgentic33.7%67.7——Artificial Analysis / Mercor
FrontierMath-Tier-4-v2-PrivateMath29.3%62.0max effort—Epoch AI
Artificial Analysis Omniscience AccuracyKnowledge24.3%43.9——Artificial Analysis
ARC-AGI-2 (semi-private)Reasoning22.8%51.2——ARC Prize Foundation
Chess PuzzlesReasoning21.0%50.7max effort—Epoch AI
Critical Physics TasksReasoning20.9%82.2——Artificial Analysis
ResearchClawBenchAgentic20.7%———InternScience
Mystery Game PuzzlesReasoning15.0%49.9medium effort—Epoch AI
EBR-benchReasoning9.5%55.8max effort—Epoch AI
Agent Arena task outcomeAgentic4.972.9max effort15 Sept 2026LMArena
Agent Arena steerabilityAgentic4.972.9max effort15 Sept 2026LMArena
Terminal-Bench 3.0Agentic4.6%58.5——Ryan Marten et al.
Agent Arena command recoveryAgentic1.969.5max effort15 Sept 2026LMArena
ProgramBenchCoding0.5%——21 Sept 2026Vals AI

48 benchmarks count, from 57 of 62 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsOpenRouter, collected directlyNo licence statedEpoch AI, collected directlyCC BY — free to use and redistribute with attributionLiveBench, collected directlyNo licence stated for the leaderboard. The site repo serving these CSVs has no LICENSE; the harness repo carries upstream Apache-2.0 and MIT copies that cover code, not resultsVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AIARC Prize Foundation, collected directlyNo licence stated. Their terms ask for written permission before commercial useSierra Research, collected directlyMIT — results are in the licensed repositoryLMArena, collected directlyCC BY 4.0 (lmarena-ai/leaderboard-dataset on Hugging Face)

Same level, lower price

Qwen3.8-Flash-Next66.4 · Freedots3-note Preview65.5 · FreeDeepSeek V4 Flash 042363.0 · $0.177

More from Z.AI

GLM-5.369.6GLM-5.3-Flash65.6GLM-5.157.8GLM-552.2GLM-4.750.2