Model comparison
GLM-5.2 vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 51.1 on the Noometry Index. GLM-5.2 costs 9.3× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 42 shared benchmarks.
Summary
- They share 42 benchmarks with published results for both. GLM-5.2 scores higher in 3 categories and GPT-6 Astra in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 42.3.
- The biggest single-benchmark swing is ARC-AGI-2: 22.8% for GLM-5.2 and 95% for GPT-6 Astra.
- GLM-5.2 is cheaper at $1.40 / $4.40 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra 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-6 Astra | |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 51.1 | 70.8 |
| Released | 2026-06-13 | 2026-09-03 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 131K | 128K |
| Input $ / M tokens | $1.40 | $10 |
| Output $ / M tokens | $4.40 | $50 |
| Results tracked | 51 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
GLM-5.2: 51.3 (#41), GPT-6 Astra: 73.7 (#2)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| DeepSWE | 43.8% | 74.1% |
| FrontierCode | 24.5% | 53.3% |
| LMArena WebDev | 1603 | 1786 |
| SciCode | 50.5% | 56.5% |
| WeirdML | 70.1% | 93.6% |
| LMArena Coding | 1485 | 1487 |
| ALE-Bench | 1,047 | 2,951 |
| SWE-bench Verified | 78.7% | — |
| FrontierSWE | — | 65.5% |
| GSO | — | 79.4% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use GPT-6 Astra leads
GLM-5.2: 32.4 (#63), GPT-6 Astra: 52.9 (#3)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| APEX-Agents | 45.2% | 64.7% |
| Vending-Bench 2 | 8,314 | 15,515 |
| Remote Labor Index | — | 20.8% |
| τ²-bench Banking | 37.1% | — |
| PostTrainBench | 31.7% | — |
| BALROG | — | 68.3% |
| GBAEval | 0% | — |
| GDP.pdf | — | 34.2% |
Reasoning GPT-6 Astra leads
GLM-5.2: 42.3 (#52), GPT-6 Astra: 85.1 (#1)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 22.8% | 95% |
| NYT Connections (extended) | 74.3% | 98.1% |
| ARC-AGI-1 | 77% | 98.5% |
| CritPt | 20.9% | 31.7% |
| Chess Puzzles | 21% | 72% |
| EBR-Bench | 9.5% | 76.2% |
| LMArena Hard Prompts | 1480 | 1462 |
| Mystery Game Puzzles | 19% | 84% |
| DTBench | 93.6% | 97.3% |
| LMCA | 45.8% | 64.4% |
| Epoch Capabilities Index | 151.78 | 166.45 |
| SimpleBench | 58.8% | — |
| Kagi LLM Benchmark | 62.6% | — |
| Surface Evolver Bench | 55.6% | — |
| Bench to the Future 3 | — | 0.14 |
Math GPT-6 Astra leads
GLM-5.2: 55.7 (#43), GPT-6 Astra: 93.5 (#2)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 59.2% | 93.7% |
| FrontierMath Tier 4 | 29.3% | 97.6% |
| OTIS Mock AIME 2024-2025 | 86.4% | 100% |
| ProofBench | 35% | 99% |
| LMArena Math | 1482 | 1465 |
| MathArena Final-Answer Competitions | 67.6% | — |
| FrontierMath Erdős | — | 2.9% |
Knowledge GPT-6 Astra leads
GLM-5.2: 57.1 (#40), GPT-6 Astra: 75.3 (#1)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 91.9% | 95.8% |
| SimpleQA Verified | 34.2% | 75.6% |
| LMArena Expert | 1486 | 1483 |
| Humanity's Last Exam | — | 54.8% |
| Vectara Hallucination Rate | — | 8.7% |
Multimodal Not comparable
GLM-5.2: —, GPT-6 Astra: 55.0 (#3)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| LMArena Vision | — | 1281 |
| Blueprint-Bench 2 | — | 49.7% |
| Furniture Assembly | — | 80% |
| LMArena Document | — | 1468 |
Multilingual GLM-5.2 leads
GLM-5.2: 55.8 (#26), GPT-6 Astra: 53.7 (#61)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1459 | 1430 |
| LMArena Chinese | 1519 | 1484 |
| LMArena French | 1479 | 1456 |
| LMArena German | 1468 | 1440 |
| LMArena Japanese | 1451 | 1379 |
| LMArena Korean | 1445 | 1426 |
| LMArena Russian | 1466 | 1436 |
| LMArena Spanish | 1477 | 1407 |
Instruction Following Too close to call
GLM-5.2: 76.9 (#34), GPT-6 Astra: 76.3 (#44)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1465 | 1450 |
Long Context Too close to call
GLM-5.2: 45.3 (#43), GPT-6 Astra: 44.5 (#62)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1479 | 1456 |
Writing & Preference GPT-6 Astra leads
GLM-5.2: 70.4 (#21), GPT-6 Astra: 75.3 (#7)
| Benchmark | GLM-5.2 | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1470 | 1441 |
| LMArena Creative Writing | 1462 | 1418 |
| EQ-Bench Creative Writing | 1757 | 2173 |
| LMArena Multi-Turn | 1469 | 1448 |
| EQ-Bench 4 | 1222 | — |
Frequently asked questions
Is GLM-5.2 better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 51.1 on the Noometry Index. GLM-5.2 costs 9.3× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Which is cheaper, GLM-5.2 or GPT-6 Astra?
GLM-5.2 is cheaper. It lists at $1.40 per million input tokens and $4.40 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is GLM-5.2 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 51.3 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 1M.
How many benchmarks do GLM-5.2 and GPT-6 Astra share?
42 benchmarks have published results for both models. GLM-5.2 has 51 scored results on Noometry and GPT-6 Astra has 56.