Model comparison
Gemma 3 4B vs GPT-5 Mini
GPT-5 Mini is the stronger model overall, scoring 41.8 to 28.1 on the Noometry Index. Gemma 3 4B costs 14× less per token, which makes it the better buy when GPT-5 Mini's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
GoogleSummary
- They share 22 benchmarks with published results for both. Gemma 3 4B scores higher in 0 categories and GPT-5 Mini in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5 Mini leads 45.6 to 11.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 7.5% for Gemma 3 4B and 86.7% for GPT-5 Mini.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.25 / $2 for GPT-5 Mini.
- GPT-5 Mini accepts more context: 400K tokens versus 131K.
- Gemma 3 4B has downloadable open weights; the other is API-only.
Side by side
| Gemma 3 4B | GPT-5 Mini | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 28.1 | 41.8 |
| Released | 2025-03-12 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 131K | 400K |
| Max output | 4K | 128K |
| Input $ / M tokens | $0.04 | $0.25 |
| Output $ / M tokens | $0.08 | $2 |
| Results tracked | 22 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 Mini leads
Gemma 3 4B: 35.9 (#215), GPT-5 Mini: 40.1 (#146)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Coding | 1230 | 1406 |
| SWE-bench Verified | — | 64.7% |
| SWE-bench Verified (bash only) | — | 59.8% |
| SWE-bench Multilingual | — | 39.7% |
| SciCode | — | 39.2% |
| WeirdML | — | 52.7% |
| ALE-Bench | — | 799.77 |
| AlgoTune | — | 1.38 |
Agentic & Tool Use GPT-5 Mini leads
Gemma 3 4B: 20.9 (#142), GPT-5 Mini: 31.1 (#70)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | 55.5% |
| Terminal-Bench | — | 34.8% |
| Vending-Bench 2 | — | -31.18 |
Reasoning GPT-5 Mini leads
Gemma 3 4B: 13.2 (#335), GPT-5 Mini: 23.9 (#168)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| Kagi LLM Benchmark | 25.2% | 70.3% |
| Chess Puzzles | 0% | 30% |
| LMArena Hard Prompts | 1253 | 1380 |
| DTBench | 50.9% | 80.5% |
| LMCA | 2.8% | 34.2% |
| Epoch Capabilities Index | 116.02 | 145.52 |
| ARC-AGI-2 | — | 4.4% |
| ARC-AGI-1 | — | 54.3% |
| CritPt | — | 0% |
| EnigmaEval | — | 8.2% |
| Mystery Game Puzzles | — | 10% |
| ForecastBench | — | 61 |
Math GPT-5 Mini leads
Gemma 3 4B: 16.8 (#292), GPT-5 Mini: 46.7 (#69)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 86.7% |
| LMArena Math | 1239 | 1378 |
| FrontierMath (Tiers 1-3) | — | 46.7% |
| FrontierMath Tier 4 | — | 12.2% |
| ProofBench | — | 9% |
| Omni-MATH | — | 72.2% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 27.2% |
| FrontierMath Tier 4 (v1) | — | 6.3% |
Knowledge GPT-5 Mini leads
Gemma 3 4B: 11.8 (#299), GPT-5 Mini: 45.6 (#86)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| GPQA Diamond | 23.2% | 75% |
| Vectara Hallucination Rate | 6.4% | 12.9% |
| LMArena Expert | 1223 | 1379 |
| Humanity's Last Exam | — | 19.4% |
| SimpleQA Verified | — | 21.6% |
| MMLU-Pro | — | 83.5% |
| Confabulations | — | 13.3% |
| GPQA (HELM) | — | 75.6% |
Multimodal Not comparable
Gemma 3 4B: —, GPT-5 Mini: 35.6 (#85)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Vision | — | 1202 |
| VPCT | — | 40.2% |
Multilingual GPT-5 Mini leads
Gemma 3 4B: 42.5 (#194), GPT-5 Mini: 48.9 (#137)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Non-English | 1273 | 1363 |
| LMArena German | 1281 | 1366 |
| LMArena Russian | 1294 | 1362 |
| LMArena Chinese | — | 1385 |
| LMArena French | — | 1386 |
| LMArena Japanese | — | 1341 |
| LMArena Korean | — | 1308 |
| LMArena Spanish | — | 1355 |
Instruction Following GPT-5 Mini leads
Gemma 3 4B: 65.2 (#225), GPT-5 Mini: 76.2 (#46)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Instruction Following | 1239 | 1357 |
| IFEval | — | 92.7% |
Long Context GPT-5 Mini leads
Gemma 3 4B: 38.7 (#194), GPT-5 Mini: 41.9 (#132)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Longer Query | 1273 | 1355 |
| Fiction.LiveBench | — | 69.4% |
Writing & Preference GPT-5 Mini leads
Gemma 3 4B: 42.0 (#239), GPT-5 Mini: 55.2 (#148)
| Benchmark | Gemma 3 4B | GPT-5 Mini |
|---|---|---|
| LMArena Text | 1291 | 1373 |
| LMArena Creative Writing | 1271 | 1325 |
| EQ-Bench Creative Writing | 1068 | 1313 |
| LMArena Multi-Turn | 1255 | 1363 |
| Short-Story Creative Writing | — | 83.1% |
| WildBench | — | 85.5% |
Frequently asked questions
Is Gemma 3 4B better than GPT-5 Mini?
GPT-5 Mini is the stronger model overall, scoring 41.8 to 28.1 on the Noometry Index. Gemma 3 4B costs 14× less per token, which makes it the better buy when GPT-5 Mini's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or GPT-5 Mini?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; GPT-5 Mini lists at $0.25 and $2.
Is Gemma 3 4B or GPT-5 Mini better for coding?
GPT-5 Mini scores higher on coding benchmarks: 40.1 versus 35.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Mini does, with 400K tokens against 131K.
How many benchmarks do Gemma 3 4B and GPT-5 Mini share?
22 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and GPT-5 Mini has 60.