Up to four models
GPT-6.1 Sol vs GLM-5.3
On the index (GPT-4 = 100), GPT-6.1 Sol scores 196 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-6.1 Sol
196180–212
WeightsClosed
ReleasedSep 2026
Price in / out$2.0 / $10
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-6.1 Sol leads GLM-5.3 by 24 points, ahead on 4 of 5 shared core tests, at 9.8× the price.
Test by testthird-party results
GPT-6.1 SolGLM-5.3
GLM-5.3Every result as a table
| Test | GPT-6.1 Sol | GLM-5.3 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 95.4% | 90.9% |
| LiveBench Data Analysis | 82.7% | 70.2% |
| LiveBench IF | 74.2% | 69.3% |
| LiveBench Language | 90.1% | 79.9% |
| LiveBench Reasoning | 92.6% | 85.8% |
| SimpleQA Verified | 73.9% | 41.0% |
| Maths | ||
| FrontierMath (tiers 1-3) | 93.7% | 68.8% |
| AIME-style maths (OTIS mock) | 100.0% | 91.1% |
| LiveBench Mathematics | 96.8% | 87.9% |
| ProofBench v1.1 | 99.0% | 49.0% |
| Chess puzzles | 68.0% | 37.0% |
| FrontierMath (tier 4) | 100.0% | 29.3% |
| Coding | ||
| Terminal-Bench 4.0 | 55.0% | 38.9% |
| LiveBench Agentic Coding | 56.8% | 60.9% |
| LiveBench Coding | 80.7% | 79.0% |
| Code Migration | 65.1% | 44.2% |
| CyberBench v1.1 | 39.3% | 72.1% |
| IOI | 96.9% | 68.4% |
| ProgramBench | 3.0% | 1.5% |
| Vibe Code Bench v1.1 | 88.9% | 78.1% |
| Agents | ||
| APEX-Agents | 60.0% | 56.6% |
| SRE Bench | 50.8% | 9.2% |
| Terminal-Bench-Science | 52.9% | 5.7% |
| DTBench | 96.5% | 87.7% |
| LMCA | 62.8% | 55.5% |
| Mystery game puzzles | 80.0% | 33.0% |
| Professional work | ||
| Finance Agent (v2) | 52.0% | 55.8% |
| EMB | 70.8% | 56.3% |
| HLAB-Vals-1.1 | 6.3% | 14.6% |
| Legal Research Bench | 38.5% | 49.0% |
| MedCode | 48.8% | 42.9% |
| MedScribe | 86.5% | 88.8% |
| Public Benefits Bench v1.1 | 59.3% | 68.5% |
| Tax Agent Bench | 21.8% | 39.7% |
| Novel problems | ||
| MysteryMechanism | 46.4% | 23.0% |
Tests at least two of these models have taken. Bold rows are the core tests.
