Up to four models
GLM-5.3 vs Gemini 3.5 Flash-Lite
On the index (GPT-4 = 100), GLM-5.3 scores 168 and Gemini 3.5 Flash-Lite 141. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
168152–184
WeightsOpen
ReleasedAug 2026
Price in / out$0.070 / $1.4
Context1.05M
Core tests taken5 of 9
141125–156
WeightsClosed
ReleasedJul 2026
Price in / out$0.30 / $2.5
Context1.05M
Core tests taken5 of 9
GLM-5.3 leads Gemini 3.5 Flash-Lite by 28 points, ahead on 4 of 4 shared core tests, and costs 2.1× less.
Test by testthird-party results
Every result as a table
| Test | GLM-5.3 | Gemini 3.5 Flash-Lite |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 90.9% | 83.3% |
| GPQA Diamond (Vals AI) | 88.1% | 83.8% |
| MMLU Pro | 86.8% | 85.8% |
| Maths | ||
| FrontierMath (tiers 1-3) | 68.8% | 26.0% |
| AIME-style maths (OTIS mock) | 91.1% | 71.1% |
| Chess puzzles | 37.0% | 21.0% |
| FrontierMath (tier 4) | 29.3% | 0.0% |
| Coding | ||
| Code Migration | 44.2% | 6.2% |
| CyberBench v1.1 | 72.1% | 59.1% |
| LiveCodeBench | 80.5% | 79.0% |
| ProgramBench | 1.5% | 0.0% |
| SWE-bench | 95.4% | 75.0% |
| Vibe Code Bench v1.1 | 78.1% | 37.2% |
| WeirdML | 75.4% | 39.0% |
| Agents | ||
| APEX-Agents | 56.6% | 29.3% |
| DTBench | 87.7% | 83.5% |
| LMCA | 55.5% | 37.6% |
| Mystery game puzzles | 33.0% | 19.0% |
| Professional work | ||
| Finance Agent (v2) | 55.8% | 47.4% |
| EMB | 56.3% | 43.2% |
| LegalBench | 84.8% | 84.1% |
| Legal Research Bench | 49.0% | 13.9% |
| MedCode | 42.9% | 43.5% |
| MedScribe | 88.8% | 70.9% |
| Public Benefits Bench v1.1 | 68.5% | 51.8% |
| Tax Agent Bench | 43.3% | 7.2% |
| TaxEval v2 | 72.4% | 72.6% |
Tests at least two of these models have taken. Bold rows are the core tests.