Model comparison
GLM-5.3 vs GPT-4.1
GLM-5.3 is the stronger model overall, scoring 54.8 to 35.9 on the Noometry Index.
Last verified . 28 shared benchmarks.
Summary
- They share 28 benchmarks with published results for both. GLM-5.3 scores higher in 9 categories and GPT-4.1 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.3 leads 62.3 to 22.3.
- The biggest single-benchmark swing is FrontierMath (Tiers 1-3): 68.8% for GLM-5.3 and 6% for GPT-4.1.
- GLM-5.3 is cheaper at $1.40 / $4.40 per million input/output tokens, against $2 / $8 for GPT-4.1.
- GPT-4.1 accepts more context: 1.05M tokens versus 1M.
- GLM-5.3 has downloadable open weights; the other is API-only.
Side by side
| GLM-5.3 | GPT-4.1 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 54.8 | 35.9 |
| Released | 2026-08-14 | 2025-04-14 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 131K | 33K |
| Input $ / M tokens | $1.40 | $2 |
| Output $ / M tokens | $4.40 | $8 |
| Results tracked | 42 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.3 leads
GLM-5.3: 59.5 (#14), GPT-4.1: 34.4 (#238)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| WeirdML | 75.4% | 39% |
| LMArena Coding | 1496 | 1391 |
| ALE-Bench | 1,317 | 558.1 |
| SWE-bench Verified | — | 48.5% |
| DeepSWE | 69% | — |
| FrontierCode | 40.1% | — |
| SWE-bench Verified (bash only) | — | 39.6% |
| Aider Polyglot | — | 52.4% |
| CursorBench | 42.6% | — |
| LMArena WebDev | 1622 | — |
| FrontierSWE | 30.2% | — |
| SciCode | 59% | — |
| CadEval | — | 42% |
Agentic & Tool Use GLM-5.3 leads
GLM-5.3: 36.4 (#38), GPT-4.1: 34.7 (#43)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| APEX-Agents | 56.6% | — |
| Berkeley Function Calling Leaderboard | — | 54% |
| Vending-Bench 2 | 8,164 | — |
Reasoning GLM-5.3 leads
GLM-5.3: 46.1 (#46), GPT-4.1: 11.7 (#339)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| Chess Puzzles | 21% | 6% |
| LMArena Hard Prompts | 1489 | 1384 |
| DTBench | 87.7% | 68.3% |
| LMCA | 55.5% | 25.6% |
| Epoch Capabilities Index | 155.61 | 136.78 |
| ARC-AGI-2 | — | 0.4% |
| SimpleBench | — | 27% |
| Kagi LLM Benchmark | — | 52.3% |
| NYT Connections (extended) | 74.2% | — |
| ARC-AGI-1 | — | 5.5% |
| CritPt | 19.1% | — |
| EnigmaEval | — | 2.2% |
| Mystery Game Puzzles | 33% | — |
| Bench to the Future 3 | 0.15 | — |
| ForecastBench | — | 61.5 |
Math GLM-5.3 leads
GLM-5.3: 62.3 (#33), GPT-4.1: 22.3 (#280)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 68.8% | 6% |
| OTIS Mock AIME 2024-2025 | 91.1% | 38.3% |
| LMArena Math | 1489 | 1370 |
| FrontierMath Tier 4 | 29.3% | — |
| ProofBench | 49% | — |
| Omni-MATH | — | 47.1% |
| MATH Level 5 | — | 83% |
| FrontierMath (Feb 2025 set) | — | 5.5% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge GLM-5.3 leads
GLM-5.3: 58.3 (#37), GPT-4.1: 37.1 (#160)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| GPQA Diamond | 90.9% | 66.9% |
| SimpleQA Verified | 41% | 31.1% |
| LMArena Expert | 1516 | 1364 |
| Humanity's Last Exam | — | 5.4% |
| MMLU-Pro | — | 81.1% |
| Vectara Hallucination Rate | — | 5.6% |
| GPQA (HELM) | — | 65.9% |
Multimodal Not comparable
GLM-5.3: —, GPT-4.1: 38.2 (#67)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| LMArena Vision | — | 1211 |
| GeoBench | — | 72% |
Multilingual GLM-5.3 leads
GLM-5.3: 55.7 (#28), GPT-4.1: 49.4 (#133)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| LMArena Non-English | 1457 | 1370 |
| LMArena Chinese | 1528 | 1382 |
| LMArena French | 1499 | 1382 |
| LMArena German | 1499 | 1381 |
| LMArena Japanese | 1453 | 1319 |
| LMArena Korean | 1472 | 1339 |
| LMArena Russian | 1463 | 1377 |
| LMArena Spanish | 1460 | 1376 |
Instruction Following GLM-5.3 leads
GLM-5.3: 77.5 (#23), GPT-4.1: 71.3 (#153)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| LMArena Instruction Following | 1477 | 1367 |
| IFEval | — | 83.8% |
Long Context GLM-5.3 leads
GLM-5.3: 45.4 (#41), GPT-4.1: 40.0 (#163)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| LMArena Longer Query | 1482 | 1385 |
| Fiction.LiveBench | — | 63.9% |
Writing & Preference GLM-5.3 leads
GLM-5.3: 75.7 (#6), GPT-4.1: 57.6 (#125)
| Benchmark | GLM-5.3 | GPT-4.1 |
|---|---|---|
| LMArena Text | 1471 | 1383 |
| LMArena Creative Writing | 1457 | 1363 |
| EQ-Bench Creative Writing | 2075 | 1420 |
| LMArena Multi-Turn | 1472 | 1398 |
| WildBench | — | 85.4% |
Frequently asked questions
Is GLM-5.3 better than GPT-4.1?
GLM-5.3 is the stronger model overall, scoring 54.8 to 35.9 on the Noometry Index.
Which is cheaper, GLM-5.3 or GPT-4.1?
GLM-5.3 is cheaper. It lists at $1.40 per million input tokens and $4.40 per million output tokens; GPT-4.1 lists at $2 and $8.
Is GLM-5.3 or GPT-4.1 better for coding?
GLM-5.3 scores higher on coding benchmarks: 59.5 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
GPT-4.1 does, with 1.05M tokens against 1M.
How many benchmarks do GLM-5.3 and GPT-4.1 share?
28 benchmarks have published results for both models. GLM-5.3 has 42 scored results on Noometry and GPT-4.1 has 52.