HOST DIAGNOSTIC · Zhipu AI
GLM-5.1
GLM-5.1· GLM· OPEN WEIGHTS
ACTIVE
HOST IDENTIFICATION
Zhipu AI
GLM-5.1
ACTIVE
AVG SCORE
[12.1]
CAPABLE
STRONGEST
Cost Efficiency
[15]
WEAKEST
Candor
[11]
MAXED (20/20)
0
of 20 attributes
20-DIMENSION EVALUATION
ATTRIBUTE SCORES · 20 DIMENSIONS
Cognitive
Bulk Apperception
12
Reasoning
12
Mathematical Precision
13
World Knowledge
12
Scientific Acumen
13
Technical
Code Generation
14
Tool Use
11
Multimodal Fluency
11
Speed
13
Cost Efficiency
15
Behavioral
Candor
11
Creativity
12
Tenacity
12
Self-Correction
11
Calibration
11
Operational
Instruction Following
12
Context Fidelity
12
Multi-turn Coherence
12
Planning and Execution
12
Safety Alignment
11
SCORING METHODOLOGY
CALIBRATED. Added to Arena Text April 7, 2026 + Arena Code April 9. GLM-5 baseline: SWE-Bench Verified 77.8% (94.6% of Opus 4.6's 80.8%). GLM-5.1 is the upgraded variant. Cost Efficiency scored high (17) as Chinese open-weight models deliver strong performance at fraction of frontier pricing. Represents the 'open-weight models closing the gap' narrative highlighted in Stanford AI Index 2026. Provider formerly known as Zhipu AI, rebranded to Z.ai.