Model comparison
GLM-4.5-Air vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 38.7 on the Noometry Index. GLM-4.5-Air costs 11× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
Summary
- They share 22 benchmarks with published results for both. GLM-4.5-Air scores higher in 0 categories and GPT-5.2 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.2 leads 50.1 to 23.3.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 43% for GLM-4.5-Air and 73.3% for GPT-5.2.
- GLM-4.5-Air is cheaper at $0.20 / $1.10 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- GPT-5.2 accepts more context: 400K tokens versus 131K.
- GLM-4.5-Air has downloadable open weights; the other is API-only.
Side by side
| GLM-4.5-Air | GPT-5.2 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 38.7 | 54.1 |
| Released | 2025-07-20 | 2025-12-11 |
| Weights | Open | Proprietary |
| Context window | 131K | 400K |
| Max output | 98K | 128K |
| Input $ / M tokens | $0.20 | $1.75 |
| Output $ / M tokens | $1.10 | $14 |
| Results tracked | 27 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
GLM-4.5-Air: 34.5 (#225), GPT-5.2: 52.2 (#32)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| GSO | 2.9% | 27.4% |
| LMArena Coding | 1397 | 1447 |
| SWE-bench Verified | — | 73.8% |
| SWE-bench Verified (bash only) | — | 72.8% |
| LMArena WebDev | — | 1415 |
| SWE-bench Multilingual | — | 66.7% |
| WeirdML | — | 72.2% |
| ALE-Bench | — | 1,294 |
| AlgoTune | — | 2.05 |
Agentic & Tool Use Not comparable
GLM-4.5-Air: —, GPT-5.2: 40.0 (#24)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| Terminal-Bench | — | 64.9% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| Remote Labor Index | — | 2.5% |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| DeepResearch Bench | — | 41.1% |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 75.3% |
| Vending-Bench 2 | — | 3,591 |
Reasoning GPT-5.2 leads
GLM-4.5-Air: 23.3 (#166), GPT-5.2: 50.1 (#33)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| Kagi LLM Benchmark | 43% | 73.3% |
| LMArena Hard Prompts | 1379 | 1445 |
| ForecastBench | 59.2 | 60.1 |
| ARC-AGI-2 | — | 52.9% |
| SimpleBench | — | 45.8% |
| NYT Connections (extended) | — | 83.6% |
| ARC-AGI-1 | — | 86.2% |
| Chess Puzzles | — | 49% |
| EnigmaEval | — | 10.4% |
| EBR-Bench | — | 23% |
| Mystery Game Puzzles | — | 23% |
| DTBench | — | 90.9% |
| LMCA | — | 43.9% |
| Epoch Capabilities Index | — | 153.45 |
Math GPT-5.2 leads
GLM-4.5-Air: 33.8 (#166), GPT-5.2: 59.8 (#32)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Math | 1396 | 1439 |
| FrontierMath (Tiers 1-3) | — | 67.4% |
| FrontierMath Tier 4 | — | 31.7% |
| MathArena Final-Answer Competitions | — | 72% |
| OTIS Mock AIME 2024-2025 | — | 96.1% |
| ProofBench | — | 15% |
| Omni-MATH | 39.1% | — |
| FrontierMath (Feb 2025 set) | — | 40.7% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge GPT-5.2 leads
GLM-4.5-Air: 34.6 (#184), GPT-5.2: 59.2 (#29)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| Humanity's Last Exam | 8.1% | 27.8% |
| Vectara Hallucination Rate | 9.3% | 8.4% |
| LMArena Expert | 1370 | 1445 |
| GPQA Diamond | — | 91.4% |
| SimpleQA Verified | — | 37.1% |
| MMLU-Pro | 76.2% | — |
| GPQA (HELM) | 59.4% | — |
Multimodal Not comparable
GLM-4.5-Air: —, GPT-5.2: 50.9 (#7)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Vision | — | 1268 |
| VPCT | — | 84% |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |
Multilingual GPT-5.2 leads
GLM-4.5-Air: 49.4 (#130), GPT-5.2: 53.6 (#63)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1366 | 1425 |
| LMArena Chinese | 1427 | 1460 |
| LMArena French | 1399 | 1456 |
| LMArena German | 1376 | 1448 |
| LMArena Japanese | 1348 | 1421 |
| LMArena Korean | 1306 | 1390 |
| LMArena Russian | 1373 | 1440 |
| LMArena Spanish | 1385 | 1433 |
Instruction Following GPT-5.2 leads
GLM-4.5-Air: 69.8 (#165), GPT-5.2: 74.9 (#83)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1353 | 1417 |
| IFEval | 81.2% | — |
Long Context GPT-5.2 leads
GLM-4.5-Air: 41.6 (#130), GPT-5.2: 44.0 (#74)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1367 | 1428 |
| CL-bench | — | 18.2% |
Writing & Preference GPT-5.2 leads
GLM-4.5-Air: 56.1 (#134), GPT-5.2: 67.0 (#30)
| Benchmark | GLM-4.5-Air | GPT-5.2 |
|---|---|---|
| LMArena Text | 1384 | 1439 |
| LMArena Creative Writing | 1343 | 1402 |
| LMArena Multi-Turn | 1371 | 1458 |
| EQ-Bench Creative Writing | — | 1703 |
| WildBench | 78.9% | — |
Frequently asked questions
Is GLM-4.5-Air better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 38.7 on the Noometry Index. GLM-4.5-Air costs 11× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Which is cheaper, GLM-4.5-Air or GPT-5.2?
GLM-4.5-Air is cheaper. It lists at $0.20 per million input tokens and $1.10 per million output tokens; GPT-5.2 lists at $1.75 and $14.
Is GLM-4.5-Air or GPT-5.2 better for coding?
GPT-5.2 scores higher on coding benchmarks: 52.2 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5.2 does, with 400K tokens against 131K.
How many benchmarks do GLM-4.5-Air and GPT-5.2 share?
22 benchmarks have published results for both models. GLM-4.5-Air has 27 scored results on Noometry and GPT-5.2 has 67.