Model comparison
GLM-5.3 vs GLM-5.3-Flash
GLM-5.3 is the stronger model overall, scoring 54.8 to 51.8 on the Noometry Index. GLM-5.3-Flash costs 9.1× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
Summary
- They share 35 benchmarks with published results for both. GLM-5.3 scores higher in 4 categories and GLM-5.3-Flash in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GLM-5.3 leads 75.7 to 65.3.
- The biggest single-benchmark swing is ProofBench: 49% for GLM-5.3 and 21% for GLM-5.3-Flash.
- GLM-5.3-Flash is cheaper at $0.15 / $0.50 per million input/output tokens, against $1.40 / $4.40 for GLM-5.3.
Side by side
| GLM-5.3 | GLM-5.3-Flash | |
|---|---|---|
| Provider | Z.ai (Zhipu) | Z.ai (Zhipu) |
| Noometry Index | 54.8 | 51.8 |
| Released | 2026-08-14 | 2026-08-20 |
| Weights | Open | Open |
| Context window | 1M | 1M |
| Max output | 131K | 131K |
| Input $ / M tokens | $1.40 | $0.15 |
| Output $ / M tokens | $4.40 | $0.50 |
| Results tracked | 42 | 40 |
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), GLM-5.3-Flash: 53.1 (#31)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| DeepSWE | 69% | 63.4% |
| FrontierCode | 40.1% | 31.8% |
| CursorBench | 42.6% | 36.8% |
| LMArena WebDev | 1622 | 1609 |
| FrontierSWE | 30.2% | 18.1% |
| SciCode | 59% | 51.6% |
| LMArena Coding | 1496 | 1508 |
| ALE-Bench | 1,317 | 303.55 |
| WeirdML | 75.4% | — |
Agentic & Tool Use GLM-5.3 leads
GLM-5.3: 36.4 (#38), GLM-5.3-Flash: 34.2 (#47)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| APEX-Agents | 56.6% | 52.8% |
| GDP.pdf | — | 14% |
| Vending-Bench 2 | 8,164 | — |
Reasoning GLM-5.3-Flash leads
GLM-5.3: 46.1 (#46), GLM-5.3-Flash: 48.0 (#42)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| CritPt | 19.1% | 15.4% |
| Chess Puzzles | 21% | 14% |
| LMArena Hard Prompts | 1489 | 1491 |
| Mystery Game Puzzles | 33% | 8% |
| Bench to the Future 3 | 0.15 | 0.15 |
| Epoch Capabilities Index | 155.61 | 151.88 |
| ARC-AGI-2 | — | 65.8% |
| NYT Connections (extended) | 74.2% | — |
| ARC-AGI-1 | — | 91% |
| DTBench | 87.7% | — |
| LMCA | 55.5% | — |
| Surface Evolver Bench | — | 52.5% |
Math GLM-5.3 leads
GLM-5.3: 62.3 (#33), GLM-5.3-Flash: 53.3 (#47)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| FrontierMath (Tiers 1-3) | 68.8% | 55.8% |
| FrontierMath Tier 4 | 29.3% | 17.1% |
| OTIS Mock AIME 2024-2025 | 91.1% | 93.9% |
| ProofBench | 49% | 21% |
| LMArena Math | 1489 | 1500 |
Knowledge Too close to call
GLM-5.3: 58.3 (#37), GLM-5.3-Flash: 58.4 (#36)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| GPQA Diamond | 90.9% | 90.2% |
| LMArena Expert | 1516 | 1513 |
| SimpleQA Verified | 41% | — |
Multimodal Not comparable
GLM-5.3: —, GLM-5.3-Flash: 42.8 (#27)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| LMArena Vision | — | 1296 |
Multilingual Too close to call
GLM-5.3: 55.7 (#28), GLM-5.3-Flash: 56.0 (#25)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| LMArena Non-English | 1457 | 1462 |
| LMArena Chinese | 1528 | 1527 |
| LMArena French | 1499 | 1496 |
| LMArena German | 1499 | 1470 |
| LMArena Japanese | 1453 | 1429 |
| LMArena Korean | 1472 | 1446 |
| LMArena Russian | 1463 | 1469 |
| LMArena Spanish | 1460 | 1471 |
Instruction Following Too close to call
GLM-5.3: 77.5 (#23), GLM-5.3-Flash: 77.5 (#20)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| LMArena Instruction Following | 1477 | 1478 |
Long Context Too close to call
GLM-5.3: 45.4 (#41), GLM-5.3-Flash: 45.4 (#39)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| LMArena Longer Query | 1482 | 1482 |
Writing & Preference GLM-5.3 leads
GLM-5.3: 75.7 (#6), GLM-5.3-Flash: 65.3 (#50)
| Benchmark | GLM-5.3 | GLM-5.3-Flash |
|---|---|---|
| LMArena Text | 1471 | 1471 |
| LMArena Creative Writing | 1457 | 1442 |
| LMArena Multi-Turn | 1472 | 1467 |
| EQ-Bench Creative Writing | 2075 | — |
Frequently asked questions
Is GLM-5.3 better than GLM-5.3-Flash?
GLM-5.3 is the stronger model overall, scoring 54.8 to 51.8 on the Noometry Index. GLM-5.3-Flash costs 9.1× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.
Which is cheaper, GLM-5.3 or GLM-5.3-Flash?
GLM-5.3-Flash is cheaper. It lists at $0.15 per million input tokens and $0.50 per million output tokens; GLM-5.3 lists at $1.40 and $4.40.
Is GLM-5.3 or GLM-5.3-Flash better for coding?
GLM-5.3 scores higher on coding benchmarks: 59.5 versus 53.1 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do GLM-5.3 and GLM-5.3-Flash share?
35 benchmarks have published results for both models. GLM-5.3 has 42 scored results on Noometry and GLM-5.3-Flash has 40.