Model comparison
GLM-5.2 vs GPT-5.6 Luna
GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 50.9 on the Noometry Index.
Last verified . 44 shared benchmarks.
Summary
- They share 44 benchmarks with published results for both. GLM-5.2 scores higher in 4 categories and GPT-5.6 Luna in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.5 to 55.4.
- The biggest single-benchmark swing is ARC-AGI-2: 22.8% for GLM-5.2 and 59.5% for GPT-5.6 Luna.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $1.40 / $4.40 for GLM-5.2.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 1M.
- GLM-5.2 has downloadable open weights; the other is API-only.
Side by side
| GLM-5.2 | GPT-5.6 Luna | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 50.9 | 54.7 |
| Released | 2026-06-13 | 2026-07-09 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 131K | 128K |
| Input $ / M tokens | $1.40 | $0.20 |
| Output $ / M tokens | $4.40 | $1.20 |
| Results tracked | 50 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
GLM-5.2: 50.9 (#37), GPT-5.6 Luna: 53.9 (#25)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| DeepSWE | 43.8% | 67.2% |
| FrontierCode | 24.5% | 39.8% |
| LMArena WebDev | 1604 | 1517 |
| SciCode | 50.5% | 53.6% |
| WeirdML | 70.1% | 60.9% |
| LMArena Coding | 1484 | 1466 |
| ALE-Bench | 1,047 | 1,667 |
| SWE-bench Verified | 78.7% | — |
| CursorBench | — | 35.9% |
Agentic & Tool Use GPT-5.6 Luna leads
GLM-5.2: 31.6 (#63), GPT-5.6 Luna: 35.3 (#37)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| Vending-Bench 2 | 8,314 | 4,095 |
| τ²-bench Banking | 37.1% | — |
| PostTrainBench | 31.7% | — |
| BALROG | — | 45.6% |
| GBAEval | 0% | — |
| GDP.pdf | — | 22.7% |
Reasoning GPT-5.6 Luna leads
GLM-5.2: 42.1 (#46), GPT-5.6 Luna: 47.5 (#38)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| ARC-AGI-2 | 22.8% | 59.5% |
| SimpleBench | 58.8% | 46.8% |
| Kagi LLM Benchmark | 62.6% | 49.1% |
| NYT Connections (extended) | 74.3% | 69.4% |
| ARC-AGI-1 | 77% | 88% |
| CritPt | 20.9% | 20.6% |
| Chess Puzzles | 21% | 40% |
| LMArena Hard Prompts | 1480 | 1452 |
| Mystery Game Puzzles | 19% | 21% |
| DTBench | 93.6% | 89.1% |
| LMCA | 45.8% | 48.5% |
| Surface Evolver Bench | 55.6% | 61.9% |
| Epoch Capabilities Index | 151.76 | 156.32 |
| EBR-Bench | 9.5% | — |
Math GPT-5.6 Luna leads
GLM-5.2: 55.4 (#35), GPT-5.6 Luna: 77.5 (#12)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 59.2% | 82.1% |
| FrontierMath Tier 4 | 29.3% | 61% |
| OTIS Mock AIME 2024-2025 | 86.4% | 98.3% |
| ProofBench | 35% | 60% |
| LMArena Math | 1482 | 1455 |
| MathArena Final-Answer Competitions | 67.6% | — |
Knowledge GPT-5.6 Luna leads
GLM-5.2: 57.1 (#36), GPT-5.6 Luna: 58.6 (#31)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 91.9% | 91.6% |
| SimpleQA Verified | 34.2% | 41% |
| LMArena Expert | 1484 | 1479 |
Multimodal Not comparable
GLM-5.2: —, GPT-5.6 Luna: 43.0 (#22)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | — | 1259 |
| Blueprint-Bench 2 | — | 22.6% |
| Furniture Assembly | — | 42.5% |
| LMArena Document | — | 1457 |
Multilingual GLM-5.2 leads
GLM-5.2: 56.0 (#25), GPT-5.6 Luna: 53.0 (#73)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1459 | 1418 |
| LMArena Chinese | 1518 | 1469 |
| LMArena French | 1476 | 1457 |
| LMArena German | 1470 | 1459 |
| LMArena Japanese | 1451 | 1416 |
| LMArena Korean | 1445 | 1414 |
| LMArena Russian | 1467 | 1428 |
| LMArena Spanish | 1477 | 1445 |
Instruction Following GLM-5.2 leads
GLM-5.2: 77.1 (#32), GPT-5.6 Luna: 76.0 (#48)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1465 | 1440 |
Long Context GLM-5.2 leads
GLM-5.2: 45.3 (#43), GPT-5.6 Luna: 43.9 (#78)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1479 | 1437 |
Writing & Preference GLM-5.2 leads
GLM-5.2: 70.7 (#20), GPT-5.6 Luna: 68.3 (#27)
| Benchmark | GLM-5.2 | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1470 | 1432 |
| LMArena Creative Writing | 1465 | 1397 |
| EQ-Bench Creative Writing | 1757 | 1829 |
| EQ-Bench 4 | 1222 | 1156 |
| LMArena Multi-Turn | 1468 | 1433 |
Frequently asked questions
Is GLM-5.2 better than GPT-5.6 Luna?
GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 50.9 on the Noometry Index.
Which is cheaper, GLM-5.2 or GPT-5.6 Luna?
GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; GLM-5.2 lists at $1.40 and $4.40.
Is GLM-5.2 or GPT-5.6 Luna better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 53.9 versus 50.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 1M.
How many benchmarks do GLM-5.2 and GPT-5.6 Luna share?
44 benchmarks have published results for both models. GLM-5.2 has 50 scored results on Noometry and GPT-5.6 Luna has 51.