Up to four models
GLM-5.3-Flash vs Qwen3.7-Max
On the index (GPT-4 = 100), GLM-5.3-Flash scores 161 and Qwen3.7-Max 160. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
161145–177
WeightsOpen
ReleasedAug 2026
Price in / out$0.15 / $0.50
Context1.05M
Core tests taken6 of 9
160144–177
WeightsClosed
ReleasedMay 2026
Price in / out$1.48 / $4.42
Context1M
Core tests taken4 of 9
GLM-5.3-Flash and Qwen3.7-Max are too close to call (1 point apart, within the uncertainty), ahead on 1 of 3 shared core tests, and costs 9.3× less.
Test by testthird-party results
Every result as a table
| Test | GLM-5.3-Flash | Qwen3.7-Max |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 90.1% | 90.9% |
| LiveBench Data Analysis | 76.4% | 71.8% |
| LiveBench IF | 52.8% | 74.0% |
| LiveBench Language | 77.3% | 79.7% |
| LiveBench Reasoning | 77.6% | 83.3% |
| GPQA Diamond (Vals AI) | 86.4% | 90.1% |
| MMLU Pro | 86.1% | 89.3% |
| Maths | ||
| FrontierMath (tiers 1-3) | 55.8% | 64.6% |
| AIME-style maths (OTIS mock) | 93.9% | 95.6% |
| LiveBench Mathematics | 81.2% | 85.3% |
| Chess puzzles | 21.0% | 27.0% |
| FrontierMath (tier 4) | 17.1% | 34.2% |
| Coding | ||
| LiveBench Agentic Coding | 56.8% | 43.6% |
| LiveBench Coding | 79.0% | 74.2% |
| Code Migration | 20.5% | 13.1% |
| LiveCodeBench | 80.5% | 87.1% |
| SWE-bench | 92.0% | 68.8% |
| Vibe Code Bench v1.1 | 30.8% | 47.7% |
| Agents | ||
| Mystery game puzzles | 8.0% | 32.0% |
| Professional work | ||
| Finance Agent (v2) | 57.9% | 47.8% |
| EMB | 55.9% | 57.0% |
| LegalBench | 83.9% | 84.9% |
| Legal Research Bench | 45.2% | 25.5% |
| MedScribe | 88.9% | 79.4% |
| Tax Agent Bench | 31.6% | 15.8% |
| TaxEval v2 | 75.6% | 75.3% |
Tests at least two of these models have taken. Bold rows are the core tests.