Up to four models
Gemini 3.5 Flash-Lite vs GPT-5 mini
On the index (GPT-4 = 100), Gemini 3.5 Flash-Lite scores 141 and GPT-5 mini 140. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
141125–156
WeightsClosed
ReleasedJul 2026
Price in / out$0.30 / $2.5
Context1.05M
Core tests taken5 of 9
GPT-5 mini
140124–156
WeightsClosed
ReleasedAug 2025
Price in / out$0.25 / $2.0
Context400K
Core tests taken5 of 9
Gemini 3.5 Flash-Lite and GPT-5 mini are too close to call (1 point apart, within the uncertainty), ahead on 2 of 3 shared core tests, at a similar price.
Test by testthird-party results
Every result as a table
| Test | Gemini 3.5 Flash-Lite | GPT-5 mini |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 83.3% | 75.0% |
| GPQA Diamond (Vals AI) | 83.8% | 80.3% |
| MMLU Pro | 85.8% | 82.2% |
| MMMU Pro | 83.6% | 78.9% |
| Maths | ||
| FrontierMath (tiers 1-3) | 26.0% | 46.7% |
| AIME-style maths (OTIS mock) | 71.1% | 86.7% |
| Chess puzzles | 21.0% | 24.0% |
| FrontierMath (tier 4) | 0.0% | 12.2% |
| Coding | ||
| LiveCodeBench | 79.0% | 86.6% |
| SWE-bench | 75.0% | 60.8% |
| Vibe Code Bench v1.1 | 37.2% | 14.2% |
| WeirdML | 39.0% | 52.7% |
| Agents | ||
| DTBench | 83.5% | 80.5% |
| LMCA | 37.6% | 34.2% |
| Mystery game puzzles | 19.0% | 10.0% |
| Professional work | ||
| CorpFin v2 | 60.7% | 60.2% |
| LegalBench | 84.1% | 81.8% |
| MedCode | 43.5% | 43.0% |
| MedScribe | 70.9% | 80.6% |
| MortgageTax | 68.7% | 66.9% |
| TaxEval v2 | 72.6% | 75.2% |
| Novel problems | ||
| ARC-AGI-2 | 10.3% | 4.4% |
| ARC-AGI-1 | 53.5% | 54.3% |
Tests at least two of these models have taken. Bold rows are the core tests.