Model comparison
Gemini 3.1 Flash Lite vs GPT-5.4 mini
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 40.4 on the Noometry Index. Gemini 3.1 Flash Lite costs 3.0× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Last verified . 36 shared benchmarks.
GoogleSummary
- They share 36 benchmarks with published results for both. Gemini 3.1 Flash Lite scores higher in 2 categories and GPT-5.4 mini in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.4 mini leads 51.4 to 41.7.
- The biggest single-benchmark swing is NYT Connections (extended): 8.2% for Gemini 3.1 Flash Lite and 61.8% for GPT-5.4 mini.
- Gemini 3.1 Flash Lite is cheaper at $0.25 / $1.50 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- Gemini 3.1 Flash Lite accepts more context: 1.05M tokens versus 400K.
Side by side
| Gemini 3.1 Flash Lite | GPT-5.4 mini | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 40.4 | 44.8 |
| Released | 2026-03-03 | 2026-03-17 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 400K |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.25 | $0.75 |
| Output $ / M tokens | $1.50 | $4.50 |
| Results tracked | 38 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 37.5 (#179), GPT-5.4 mini: 45.0 (#67)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena WebDev | 1256 | 1397 |
| SciCode | 41.9% | 49.9% |
| WeirdML | 52.2% | 60.3% |
| LMArena Coding | 1400 | 1438 |
| ALE-Bench | 797.73 | 1,189 |
| FrontierCode | — | 27% |
Agentic & Tool Use Too close to call
Gemini 3.1 Flash Lite: 29.9 (#72), GPT-5.4 mini: 29.6 (#74)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| DeepResearch Bench | 37.3% | 36.3% |
Reasoning GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 22.3 (#181), GPT-5.4 mini: 30.3 (#77)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| Kagi LLM Benchmark | 67.2% | 37.9% |
| NYT Connections (extended) | 8.2% | 61.8% |
| CritPt | 1.1% | 10% |
| Chess Puzzles | 25% | 24% |
| Thematic Generalization | 63.3% | 61.7% |
| LMArena Hard Prompts | 1407 | 1424 |
| DTBench | 76.8% | 80% |
| LMCA | 35% | 40.8% |
| Epoch Capabilities Index | 144.45 | 148.83 |
| ForecastBench | 54.4 | 57 |
| ARC-AGI-2 | — | 18.9% |
| ARC-AGI-1 | — | 63.7% |
| EnigmaEval | 3% | — |
| Mystery Game Puzzles | — | 11% |
Math GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 39.5 (#82), GPT-5.4 mini: 45.4 (#63)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 27.7% | 51.2% |
| OTIS Mock AIME 2024-2025 | 80% | 88.9% |
| LMArena Math | 1427 | 1418 |
| FrontierMath Tier 4 | — | 9.8% |
| ProofBench | — | 21% |
| FrontierMath (Feb 2025 set) | — | 28.3% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 41.7 (#95), GPT-5.4 mini: 51.4 (#62)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| GPQA Diamond | 81.8% | 86.9% |
| Vectara Hallucination Rate | 8.2% | 5.5% |
| LMArena Expert | 1397 | 1434 |
| Humanity's Last Exam | 8.6% | — |
| SimpleQA Verified | — | 29.4% |
Multimodal Too close to call
Gemini 3.1 Flash Lite: 38.7 (#57), GPT-5.4 mini: 39.0 (#51)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena Vision | 1240 | 1245 |
Multilingual Too close to call
Gemini 3.1 Flash Lite: 52.5 (#82), GPT-5.4 mini: 52.1 (#94)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena Non-English | 1410 | 1405 |
| LMArena Chinese | 1461 | 1446 |
| LMArena French | 1423 | 1440 |
| LMArena German | 1429 | 1407 |
| LMArena Japanese | 1411 | 1375 |
| LMArena Korean | 1393 | 1369 |
| LMArena Russian | 1420 | 1417 |
| LMArena Spanish | 1420 | 1403 |
Instruction Following GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 72.9 (#126), GPT-5.4 mini: 74.3 (#99)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena Instruction Following | 1376 | 1405 |
Long Context Too close to call
Gemini 3.1 Flash Lite: 42.5 (#117), GPT-5.4 mini: 42.9 (#107)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena Longer Query | 1394 | 1407 |
Writing & Preference GPT-5.4 mini leads
Gemini 3.1 Flash Lite: 61.0 (#90), GPT-5.4 mini: 64.2 (#56)
| Benchmark | Gemini 3.1 Flash Lite | GPT-5.4 mini |
|---|---|---|
| LMArena Text | 1416 | 1411 |
| LMArena Creative Writing | 1401 | 1370 |
| LMArena Multi-Turn | 1417 | 1429 |
| EQ-Bench Creative Writing | — | 1665 |
Frequently asked questions
Is Gemini 3.1 Flash Lite better than GPT-5.4 mini?
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 40.4 on the Noometry Index. Gemini 3.1 Flash Lite costs 3.0× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Which is cheaper, Gemini 3.1 Flash Lite or GPT-5.4 mini?
Gemini 3.1 Flash Lite is cheaper. It lists at $0.25 per million input tokens and $1.50 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is Gemini 3.1 Flash Lite or GPT-5.4 mini better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.0 versus 37.5 in the Noometry coding category.
Which has the bigger context window?
Gemini 3.1 Flash Lite does, with 1.05M tokens against 400K.
How many benchmarks do Gemini 3.1 Flash Lite and GPT-5.4 mini share?
36 benchmarks have published results for both models. Gemini 3.1 Flash Lite has 38 scored results on Noometry and GPT-5.4 mini has 46.