Model comparison
GLM-5.3-Flash vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 51.8 on the Noometry Index. GLM-5.3-Flash costs 20× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
Summary
- They share 30 benchmarks with published results for both. GLM-5.3-Flash scores higher in 4 categories and GPT-5.2 in 6 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where GPT-5.2 leads 51.3 to 42.8.
- The biggest single-benchmark swing is Chess Puzzles: 14% for GLM-5.3-Flash and 49% for GPT-5.2.
- GLM-5.3-Flash is cheaper at $0.15 / $0.50 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- GLM-5.3-Flash accepts more context: 1M tokens versus 400K.
- GLM-5.3-Flash has downloadable open weights; the other is API-only.
Side by side
| GLM-5.3-Flash | GPT-5.2 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 51.8 | 54.1 |
| Released | 2026-08-20 | 2025-12-11 |
| Weights | Open | Proprietary |
| Context window | 1M | 400K |
| Max output | 131K | 128K |
| Input $ / M tokens | $0.15 | $1.75 |
| Output $ / M tokens | $0.50 | $14 |
| Results tracked | 40 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.3-Flash leads
GLM-5.3-Flash: 53.1 (#31), GPT-5.2: 51.6 (#37)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena WebDev | 1609 | 1416 |
| LMArena Coding | 1508 | 1447 |
| ALE-Bench | 303.55 | 1,294 |
| SWE-bench Verified | — | 73.8% |
| DeepSWE | 63.4% | — |
| FrontierCode | 31.8% | — |
| SWE-bench Verified (bash only) | — | 72.8% |
| CursorBench | 36.8% | — |
| SWE-bench Multilingual | — | 66.7% |
| FrontierSWE | 18.1% | — |
| SciCode | 51.6% | — |
| GSO | — | 27.4% |
| WeirdML | — | 72.2% |
| AlgoTune | — | 2.05 |
Agentic & Tool Use GPT-5.2 leads
GLM-5.3-Flash: 34.2 (#47), GPT-5.2: 40.2 (#24)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| Terminal-Bench | — | 64.9% |
| APEX-Agents | 52.8% | — |
| 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% |
| GDP.pdf | 14% | — |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 75.3% |
| Vending-Bench 2 | — | 3,591 |
Reasoning GPT-5.2 leads
GLM-5.3-Flash: 48.0 (#42), GPT-5.2: 50.2 (#35)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 65.8% | 52.9% |
| ARC-AGI-1 | 91% | 86.2% |
| Chess Puzzles | 14% | 49% |
| LMArena Hard Prompts | 1491 | 1445 |
| Mystery Game Puzzles | 8% | 23% |
| Epoch Capabilities Index | 151.88 | 153.45 |
| SimpleBench | — | 45.8% |
| Kagi LLM Benchmark | — | 73.3% |
| NYT Connections (extended) | — | 83.6% |
| CritPt | 15.4% | — |
| EnigmaEval | — | 10.4% |
| EBR-Bench | — | 23% |
| DTBench | — | 90.9% |
| LMCA | — | 43.9% |
| Surface Evolver Bench | 52.5% | — |
| Bench to the Future 3 | 0.15 | — |
| ForecastBench | — | 60.1 |
Math GPT-5.2 leads
GLM-5.3-Flash: 53.3 (#47), GPT-5.2: 60.0 (#38)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 55.8% | 67.4% |
| FrontierMath Tier 4 | 17.1% | 31.7% |
| OTIS Mock AIME 2024-2025 | 93.9% | 96.1% |
| ProofBench | 21% | 15% |
| LMArena Math | 1500 | 1440 |
| MathArena Final-Answer Competitions | — | 72% |
| FrontierMath (Feb 2025 set) | — | 40.7% |
| FrontierMath Tier 4 (v1) | — | 18.8% |
Knowledge Too close to call
GLM-5.3-Flash: 58.4 (#36), GPT-5.2: 59.3 (#32)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 90.2% | 91.4% |
| LMArena Expert | 1513 | 1445 |
| Humanity's Last Exam | — | 27.8% |
| SimpleQA Verified | — | 37.1% |
| Vectara Hallucination Rate | — | 8.4% |
Multimodal GPT-5.2 leads
GLM-5.3-Flash: 42.8 (#27), GPT-5.2: 51.3 (#7)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena Vision | 1296 | 1268 |
| VPCT | — | 84% |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |
Multilingual GLM-5.3-Flash leads
GLM-5.3-Flash: 56.0 (#25), GPT-5.2: 53.4 (#67)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1462 | 1425 |
| LMArena Chinese | 1527 | 1460 |
| LMArena French | 1496 | 1455 |
| LMArena German | 1470 | 1448 |
| LMArena Japanese | 1429 | 1420 |
| LMArena Korean | 1446 | 1392 |
| LMArena Russian | 1469 | 1440 |
| LMArena Spanish | 1471 | 1433 |
Instruction Following GLM-5.3-Flash leads
GLM-5.3-Flash: 77.5 (#20), GPT-5.2: 74.7 (#89)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1478 | 1417 |
Long Context GLM-5.3-Flash leads
GLM-5.3-Flash: 45.4 (#39), GPT-5.2: 44.0 (#78)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1482 | 1428 |
| CL-bench | — | 18.2% |
Writing & Preference GPT-5.2 leads
GLM-5.3-Flash: 65.3 (#50), GPT-5.2: 66.8 (#32)
| Benchmark | GLM-5.3-Flash | GPT-5.2 |
|---|---|---|
| LMArena Text | 1471 | 1439 |
| LMArena Creative Writing | 1442 | 1401 |
| LMArena Multi-Turn | 1467 | 1458 |
| EQ-Bench Creative Writing | — | 1703 |
Frequently asked questions
Is GLM-5.3-Flash better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 51.8 on the Noometry Index. GLM-5.3-Flash costs 20× 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-5.3-Flash or GPT-5.2?
GLM-5.3-Flash is cheaper. It lists at $0.15 per million input tokens and $0.50 per million output tokens; GPT-5.2 lists at $1.75 and $14.
Is GLM-5.3-Flash or GPT-5.2 better for coding?
GLM-5.3-Flash scores higher on coding benchmarks: 53.1 versus 51.6 in the Noometry coding category.
Which has the bigger context window?
GLM-5.3-Flash does, with 1M tokens against 400K.
How many benchmarks do GLM-5.3-Flash and GPT-5.2 share?
30 benchmarks have published results for both models. GLM-5.3-Flash has 40 scored results on Noometry and GPT-5.2 has 67.