Compare · ModelsLive · 2 picked · head to head
Claude Sonnet 4.5 vs GLM 5.2
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GLM 5.2 wins on 17/17 benchmarks
GLM 5.2 wins 17 of 17 shared benchmarks. Leads in reasoning · knowledge · general.
Category leads
reasoning·GLM 5.2knowledge·GLM 5.2general·GLM 5.2math·GLM 5.2coding·GLM 5.2
Hype vs Reality
Attention vs performance
Claude Sonnet 4.5
#221 by perf·#11 by attention
GLM 5.2
#75 by perf·#3 by attention
Best value
GLM 5.2
3.9x better value than Claude Sonnet 4.5
Claude Sonnet 4.5
4.2 pts/$
$9.00/M
GLM 5.2
16.5 pts/$
$3.40/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
z-ai
private · undisclosed
Head to head
17 benchmarks · 2 models
Claude Sonnet 4.5GLM 5.2
ARC-AGI
GLM 5.2 leads by +13.3
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Claude Sonnet 4.5
63.7
GLM 5.2
77.0
ARC-AGI-2
GLM 5.2 leads by +9.2
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Claude Sonnet 4.5
13.6
GLM 5.2
22.8
Chess Puzzles
GLM 5.2 leads by +12.6
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Sonnet 4.5
5.3
GLM 5.2
17.9
Dtbench
GLM 5.2 leads by +17.3
Claude Sonnet 4.5
72.0
GLM 5.2
89.3
Ebr Bench
GLM 5.2 leads by +7.1
Claude Sonnet 4.5
2.4
GLM 5.2
9.5
FrontierMath-Tier-4-v2-Private
GLM 5.2 leads by +26.8
Claude Sonnet 4.5
2.4
GLM 5.2
29.3
FrontierMath-Tiers-1-3-v2-Private
GLM 5.2 leads by +37.2
Claude Sonnet 4.5
25.1
GLM 5.2
62.3
GPQA diamond
GLM 5.2 leads by +13.2
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Sonnet 4.5
79.6
GLM 5.2
92.9
Lmca
GLM 5.2 leads by +8.3
Claude Sonnet 4.5
45.6
GLM 5.2
53.9
Mystery Game Puzzles
GLM 5.2 leads by +2.2
Claude Sonnet 4.5
8.6
GLM 5.2
10.8
OTIS Mock AIME 2024-2025
GLM 5.2 leads by +8.6
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Sonnet 4.5
77.8
GLM 5.2
86.4
PostTrainBench
GLM 5.2 leads by +21.8
Claude Sonnet 4.5
9.9
GLM 5.2
31.7
Proofbench
GLM 5.2 leads by +16.0
Claude Sonnet 4.5
19.0
GLM 5.2
35.0
SimpleBench
GLM 5.2 leads by +5.4
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
Claude Sonnet 4.5
45.2
GLM 5.2
50.6
SimpleQA Verified
GLM 5.2 leads by +3.5
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Sonnet 4.5
30.7
GLM 5.2
34.2
SWE-Bench verified
GLM 5.2 leads by +7.4
SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability.
Claude Sonnet 4.5
71.3
GLM 5.2
78.7
WeirdML
GLM 5.2 leads by +22.4
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Sonnet 4.5
47.7
GLM 5.2
70.1
Full benchmark table
| Benchmark | Claude Sonnet 4.5 | GLM 5.2 |
|---|---|---|
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 63.7 | 77.0 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 13.6 | 22.8 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 5.3 | 17.9 |
Dtbench | 72.0 | 89.3 |
Ebr Bench | 2.4 | 9.5 |
FrontierMath-Tier-4-v2-Private | 2.4 | 29.3 |
FrontierMath-Tiers-1-3-v2-Private | 25.1 | 62.3 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 79.6 | 92.9 |
Lmca | 45.6 | 53.9 |
Mystery Game Puzzles | 8.6 | 10.8 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 77.8 | 86.4 |
PostTrainBench | 9.9 | 31.7 |
Proofbench | 19.0 | 35.0 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 45.2 | 50.6 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 30.7 | 34.2 |
SWE-Bench verified SWE-bench Verified · 500 human-validated tasks from 12 real Python repositories (Django, Flask, scikit-learn, sympy, and others). Each task requires the model to produce a git patch that resolves a real GitHub issue and passes the test suite. The verified subset eliminates ambiguous tasks from the original SWE-bench. The benchmark remains the most-cited evaluation for code-generation capability. | 71.3 | 78.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 47.7 | 70.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $3.00 | $15.00 | 1.0M tokens (~500 books) | $60.00 | |
| $0.01 | $6.80 | 1.0M tokens (~524 books) | $17.05 |