Model comparison
Claude Sonnet 4 vs GLM-5.3-Flash
GLM-5.3-Flash is the stronger model overall, scoring 51.8 to 40.8 on the Noometry Index.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. Claude Sonnet 4 scores higher in 1 category and GLM-5.3-Flash in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GLM-5.3-Flash leads 48.0 to 22.9.
- The biggest single-benchmark swing is ARC-AGI-2: 5.9% for Claude Sonnet 4 and 65.8% for GLM-5.3-Flash.
- GLM-5.3-Flash is cheaper at $0.15 / $0.50 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- GLM-5.3-Flash accepts more context: 1M tokens versus 200K.
- GLM-5.3-Flash has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4 | GLM-5.3-Flash | |
|---|---|---|
| Provider | Anthropic | Z.ai (Zhipu) |
| Noometry Index | 40.8 | 51.8 |
| Released | 2025-05-22 | 2026-08-20 |
| Weights | Proprietary | Open |
| Context window | 200K | 1M |
| Max output | 64K | 131K |
| Input $ / M tokens | $3 | $0.15 |
| Output $ / M tokens | $15 | $0.50 |
| Results tracked | 58 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.3-Flash leads
Claude Sonnet 4: 43.5 (#88), GLM-5.3-Flash: 53.1 (#31)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| SciCode | 40% | 51.6% |
| LMArena Coding | 1414 | 1508 |
| ALE-Bench | 655.35 | 303.55 |
| DeepSWE | — | 63.4% |
| FrontierCode | — | 31.8% |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| CursorBench | — | 36.8% |
| LMArena WebDev | — | 1609 |
| FrontierSWE | — | 18.1% |
| GSO | 4.9% | — |
| WeirdML | 46.1% | — |
Agentic & Tool Use Claude Sonnet 4 leads
Claude Sonnet 4: 38.5 (#31), GLM-5.3-Flash: 34.2 (#47)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| APEX-Agents | — | 52.8% |
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| GDP.pdf | — | 14% |
| METR Time Horizons | 62% | — |
Reasoning GLM-5.3-Flash leads
Claude Sonnet 4: 22.9 (#187), GLM-5.3-Flash: 48.0 (#42)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| ARC-AGI-2 | 5.9% | 65.8% |
| ARC-AGI-1 | 40% | 91% |
| CritPt | 0.3% | 15.4% |
| LMArena Hard Prompts | 1372 | 1491 |
| Epoch Capabilities Index | 141.69 | 151.88 |
| SimpleBench | 45.5% | — |
| Kagi LLM Benchmark | 73% | — |
| Chess Puzzles | — | 14% |
| EnigmaEval | 3.1% | — |
| Mystery Game Puzzles | — | 8% |
| DTBench | 77.1% | — |
| LMCA | 29% | — |
| Surface Evolver Bench | — | 52.5% |
| Bench to the Future 3 | — | 0.15 |
| ForecastBench | 60.2 | — |
Math GLM-5.3-Flash leads
Claude Sonnet 4: 43.3 (#80), GLM-5.3-Flash: 53.3 (#47)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 93.9% |
| LMArena Math | 1375 | 1500 |
| FrontierMath (Tiers 1-3) | — | 55.8% |
| FrontierMath Tier 4 | — | 17.1% |
| ProofBench | — | 21% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge GLM-5.3-Flash leads
Claude Sonnet 4: 41.8 (#108), GLM-5.3-Flash: 58.4 (#36)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| GPQA Diamond | 79.2% | 90.2% |
| LMArena Expert | 1372 | 1513 |
| Humanity's Last Exam | 7.8% | — |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal GLM-5.3-Flash leads
Claude Sonnet 4: 26.2 (#121), GLM-5.3-Flash: 42.8 (#27)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| LMArena Vision | 1191 | 1296 |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual GLM-5.3-Flash leads
Claude Sonnet 4: 46.7 (#156), GLM-5.3-Flash: 56.0 (#25)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| LMArena Non-English | 1333 | 1462 |
| LMArena Chinese | 1350 | 1527 |
| LMArena French | 1363 | 1496 |
| LMArena German | 1331 | 1470 |
| LMArena Japanese | 1302 | 1429 |
| LMArena Korean | 1291 | 1446 |
| LMArena Russian | 1355 | 1469 |
| LMArena Spanish | 1357 | 1471 |
Instruction Following GLM-5.3-Flash leads
Claude Sonnet 4: 71.7 (#145), GLM-5.3-Flash: 77.5 (#20)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| LMArena Instruction Following | 1376 | 1478 |
| IFEval | 84% | — |
Long Context GLM-5.3-Flash leads
Claude Sonnet 4: 33.7 (#259), GLM-5.3-Flash: 45.4 (#39)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| LMArena Longer Query | 1398 | 1482 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference GLM-5.3-Flash leads
Claude Sonnet 4: 57.1 (#132), GLM-5.3-Flash: 65.3 (#50)
| Benchmark | Claude Sonnet 4 | GLM-5.3-Flash |
|---|---|---|
| LMArena Text | 1351 | 1471 |
| LMArena Creative Writing | 1345 | 1442 |
| LMArena Multi-Turn | 1376 | 1467 |
| Short-Story Creative Writing | 81.4% | — |
| EQ-Bench Creative Writing | 1483 | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than GLM-5.3-Flash?
GLM-5.3-Flash is the stronger model overall, scoring 51.8 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 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; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or GLM-5.3-Flash better for coding?
GLM-5.3-Flash scores higher on coding benchmarks: 53.1 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
GLM-5.3-Flash does, with 1M tokens against 200K.
How many benchmarks do Claude Sonnet 4 and GLM-5.3-Flash share?
26 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GLM-5.3-Flash has 40.