evals.report
BenchmarksLabsCompareRun guidesIn the wild
Z.aiGLM

GLM-4.6

Z.ai · GLM. Released Sep 30, 2025.

GLM-4.6 is a model from Z.ai in the GLM family, released Sep 30, 2025. evals.report tracks 18 reported GLM-4.6 benchmark scores across Berkeley Function Calling Leaderboard, SWE-bench Pro, FrontierMath, LMArena, Artificial Analysis Intelligence Index, Epoch Capabilities Index, MMLU-Pro, GDPval, and 10 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

Open18 results

Benchmark results 18

Compare this model
BenchmarkCategoryScoreMetricStatusDate
Berkeley Function Calling LeaderboardTool use72.38%accuracyOfficialSep 30, 2025Details
SWE-bench ProCoding9.67%% resolvedOfficialSep 30, 2025Details
FrontierMathReasoning3.82%accuracyOfficialSep 30, 2025Details
LMArenaChat preference1440source-defined ratingOfficialSep 30, 2025Details
Artificial Analysis Intelligence IndexReasoning30.2IndexUnverifiedSep 30, 2025Details
Epoch Capabilities IndexReasoning141.4IndexOfficialSep 30, 2025Details
MMLU-ProReasoning82.9%accuracyVerifiedSep 30, 2025Details
GDPvalAgents1029EloOfficialSep 30, 2025Details
LiveCodeBenchCoding56.1%Pass@1UnverifiedSep 30, 2025Details
SciCodeCoding38.4%accuracyUnverifiedSep 30, 2025Details
Global-MMLUReasoning85.6%accuracyUnverifiedSep 30, 2025Details
WebDev ArenaChat preference1355EloVerifiedSep 30, 2025Details
EQ-Bench Creative Writing v3Chat preference1393EloVerifiedSep 30, 2025Details
Design ArenaChat preference1220EloVerifiedSep 30, 2025Details
MCP-UniverseTool use25.97%Overall Success RateVerifiedSep 30, 2025Details
FrontierMath Tier 4Reasoning2.1%accuracyOfficialSep 30, 2025Details
Vectara Hallucination LeaderboardOther9.5%Hallucination RateOfficialSep 30, 2025Details
Vibe Code BenchCoding3.09%Overall accuracyVerifiedSep 30, 2025Details