Up to four models
GPT-5.6 Terra vs GLM-5.3
On the index (GPT-4 = 100), GPT-5.6 Terra scores 178 and GLM-5.3 172. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
GPT-5.6 Terra
178162–195
WeightsClosed
ReleasedJul 2026
Price in / out$2.0 / $12
Context1.05M
Core tests taken6 of 9
GLM-5.3172155–188
WeightsOpen
ReleasedAug 2026
Price in / out$0.077 / $1.4
Context1.05M
Core tests taken5 of 9
GPT-5.6 Terra leads GLM-5.3 by 7 points, ahead on 3 of 5 shared core tests, at 11× the price.
Test by testthird-party results
GPT-5.6 TerraGLM-5.3
GLM-5.3Every result as a table
| Test | GPT-5.6 Terra | GLM-5.3 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 93.3% | 90.9% |
| LiveBench Data Analysis | 79.3% | 70.2% |
| LiveBench IF | 64.6% | 69.3% |
| LiveBench Language | 82.9% | 79.9% |
| LiveBench Reasoning | 90.6% | 85.8% |
| GPQA Diamond (Vals AI) | 90.9% | 88.1% |
| MMLU Pro | 86.7% | 86.8% |
| SimpleQA Verified | 43.2% | 41.0% |
| Maths | ||
| FrontierMath (tiers 1-3) | 86.0% | 68.8% |
| AIME-style maths (OTIS mock) | 99.7% | 91.1% |
| LiveBench Mathematics | 94.9% | 87.9% |
| ProofBench v1.1 | 74.0% | 49.0% |
| Chess puzzles | 47.0% | 37.0% |
| FrontierMath (tier 4) | 70.7% | 29.3% |
| Coding | ||
| Terminal-Bench 4.0 | 22.7% | 38.9% |
| LiveBench Agentic Coding | 54.9% | 60.9% |
| LiveBench Coding | 78.3% | 79.0% |
| Code Migration | 47.8% | 44.2% |
| CyberBench v1.1 | 72.1% | 72.1% |
| IOI | 87.6% | 68.4% |
| LiveCodeBench | 85.9% | 80.5% |
| ProgramBench | 0.5% | 1.5% |
| SWE-bench | 95.4% | 95.4% |
| Vibe Code Bench 1-100 | 14.8% | 20.0% |
| Vibe Code Bench v1.1 | 74.6% | 78.1% |
| DeepSWE | 69.6% | 69.0% |
| WeirdML | 78.3% | 75.4% |
| Agents | ||
| APEX-Agents | 58.2% | 56.6% |
| SkillsBench | 58.9% | 47.5% |
| SRE Bench | 9.5% | 9.2% |
| Terminal-Bench-Science | 10.0% | 5.7% |
| DTBench | 93.3% | 87.7% |
| LMCA | 54.9% | 55.5% |
| Mystery game puzzles | 35.0% | 33.0% |
| Professional work | ||
| Finance Agent (v2) | 54.4% | 55.8% |
| EMB | 66.2% | 56.3% |
| LegalBench | 85.1% | 84.8% |
| Legal Research Bench | 41.3% | 49.0% |
| MedCode | 43.4% | 42.9% |
| MedScribe | 82.9% | 88.8% |
| Public Benefits Bench v1.1 | 62.4% | 68.5% |
| Tax Agent Bench | 30.0% | 39.7% |
| TaxEval v2 | 76.2% | 72.4% |
Tests at least two of these models have taken. Bold rows are the core tests.
