Model comparison
Gemini 3.8 Flash vs GPT-5.4
Gemini 3.8 Flash is the stronger model overall, scoring 61.8 to 59.4 on the Noometry Index.
Last verified . 44 shared benchmarks.
GoogleSummary
- They share 44 benchmarks with published results for both. Gemini 3.8 Flash scores higher in 6 categories and GPT-5.4 in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3.8 Flash leads 76.9 to 61.8.
- The biggest single-benchmark swing is FrontierMath Tier 4: 22% for Gemini 3.8 Flash and 49% for GPT-5.4.
- Gemini 3.8 Flash is cheaper at $0.75 / $3.75 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 1.05M.
Side by side
| Gemini 3.8 Flash | GPT-5.4 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 61.8 | 59.4 |
| Released | 2026-09-02 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.05M |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.75 | $2.50 |
| Output $ / M tokens | $3.75 | $15 |
| Results tracked | 50 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3.8 Flash leads
Gemini 3.8 Flash: 59.2 (#15), GPT-5.4: 52.6 (#33)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| DeepSWE | 73.8% | 51.8% |
| LMArena WebDev | 1584 | 1465 |
| SciCode | 56.6% | 56.6% |
| WeirdML | 84.8% | 77.7% |
| LMArena Coding | 1510 | 1497 |
| ALE-Bench | 1,270 | 1,607 |
| SWE-bench Verified | — | 76.9% |
| FrontierCode | 41.2% | — |
| CursorBench | 39.6% | — |
| FrontierSWE | 19.6% | — |
| GSO | — | 31.4% |
| MirrorCode | — | 15.6% |
| AlgoTune | — | 1.85 |
Agentic & Tool Use GPT-5.4 leads
Gemini 3.8 Flash: 41.8 (#21), GPT-5.4: 46.5 (#13)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| APEX-Agents | 64.3% | 52.4% |
| Vending-Bench 2 | 5,094 | 6,144 |
| Terminal-Bench | — | 81.8% |
| Remote Labor Index | 5.8% | — |
| τ²-bench Banking | — | 39.4% |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| GDP.pdf | 23.4% | — |
| LMArena Search | — | 1197 |
| METR Time Horizons | — | 74.3% |
Reasoning Gemini 3.8 Flash leads
Gemini 3.8 Flash: 76.9 (#5), GPT-5.4: 61.8 (#19)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 89.2% | 74% |
| NYT Connections (extended) | 97.4% | 91.3% |
| ARC-AGI-1 | 98.5% | 93.7% |
| CritPt | 18.3% | 23.4% |
| Chess Puzzles | 61% | 44% |
| LMArena Hard Prompts | 1508 | 1485 |
| Mystery Game Puzzles | 47% | 37% |
| DTBench | 95.7% | 94.4% |
| LMCA | 52.9% | 52% |
| Epoch Capabilities Index | 156.71 | 156.81 |
| Kagi LLM Benchmark | — | 63.8% |
| EnigmaEval | — | 16% |
| Thematic Generalization | — | 80% |
| EBR-Bench | — | 25.4% |
| Surface Evolver Bench | 76.9% | — |
| ForecastBench | — | 59.5 |
Math GPT-5.4 leads
Gemini 3.8 Flash: 65.3 (#28), GPT-5.4: 73.5 (#19)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 68.4% | 78.6% |
| FrontierMath Tier 4 | 22% | 49% |
| OTIS Mock AIME 2024-2025 | 98.9% | 97.8% |
| ProofBench | 48% | 56% |
| LMArena Math | 1528 | 1488 |
| MathArena Final-Answer Competitions | — | 83.1% |
| FrontierMath (Feb 2025 set) | — | 47.6% |
| FrontierMath Tier 4 (v1) | — | 27.1% |
Knowledge Gemini 3.8 Flash leads
Gemini 3.8 Flash: 74.8 (#2), GPT-5.4: 65.3 (#14)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 95.4% | 93.3% |
| Humanity's Last Exam | 44.5% | 36.2% |
| SimpleQA Verified | 69.7% | 45.1% |
| LMArena Expert | 1524 | 1507 |
| Vectara Hallucination Rate | — | 7% |
Multimodal GPT-5.4 leads
Gemini 3.8 Flash: 40.7 (#45), GPT-5.4: 43.7 (#20)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1314 | 1303 |
| Blueprint-Bench 2 | 38.6% | 27.1% |
| Furniture Assembly | 31.7% | 37.5% |
| LMArena Document | — | 1471 |
Multilingual Gemini 3.8 Flash leads
Gemini 3.8 Flash: 58.0 (#5), GPT-5.4: 56.2 (#23)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1491 | 1465 |
| LMArena Chinese | 1554 | 1519 |
| LMArena French | 1498 | 1493 |
| LMArena German | 1493 | 1472 |
| LMArena Japanese | 1502 | 1485 |
| LMArena Korean | 1459 | 1448 |
| LMArena Russian | 1515 | 1480 |
| LMArena Spanish | 1485 | 1454 |
Instruction Following Too close to call
Gemini 3.8 Flash: 78.0 (#13), GPT-5.4: 77.1 (#27)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1490 | 1469 |
Long Context GPT-5.4 leads
Gemini 3.8 Flash: 46.3 (#24), GPT-5.4: 50.3 (#8)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1508 | 1473 |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |
Writing & Preference Too close to call
Gemini 3.8 Flash: 72.2 (#15), GPT-5.4: 71.9 (#17)
| Benchmark | Gemini 3.8 Flash | GPT-5.4 |
|---|---|---|
| LMArena Text | 1499 | 1469 |
| LMArena Creative Writing | 1492 | 1439 |
| EQ-Bench Creative Writing | 1748 | 1840 |
| LMArena Multi-Turn | 1501 | 1482 |
| EQ-Bench 4 | — | 1272 |
Frequently asked questions
Is Gemini 3.8 Flash better than GPT-5.4?
Gemini 3.8 Flash is the stronger model overall, scoring 61.8 to 59.4 on the Noometry Index.
Which is cheaper, Gemini 3.8 Flash or GPT-5.4?
Gemini 3.8 Flash is cheaper. It lists at $0.75 per million input tokens and $3.75 per million output tokens; GPT-5.4 lists at $2.50 and $15.
Is Gemini 3.8 Flash or GPT-5.4 better for coding?
Gemini 3.8 Flash scores higher on coding benchmarks: 59.2 versus 52.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 1.05M.
How many benchmarks do Gemini 3.8 Flash and GPT-5.4 share?
44 benchmarks have published results for both models. Gemini 3.8 Flash has 50 scored results on Noometry and GPT-5.4 has 68.