Model comparison
Gemini 2.5 Flash vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 39.3 on the Noometry Index. Gemini 2.5 Flash costs 4.0× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 50 shared benchmarks.
GoogleSummary
- They share 50 benchmarks with published results for both. Gemini 2.5 Flash scores higher in 2 categories and GPT-5 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 47.5.
- The biggest single-benchmark swing is GPQA (HELM): 39% for Gemini 2.5 Flash and 79.2% for GPT-5.
- Gemini 2.5 Flash is cheaper at $0.30 / $2.50 per million input/output tokens, against $1.25 / $10 for GPT-5.
- Gemini 2.5 Flash accepts more context: 1.05M tokens versus 400K.
Side by side
| Gemini 2.5 Flash | GPT-5 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 39.3 | 50.9 |
| Released | 2025-04-17 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 400K |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.30 | $1.25 |
| Output $ / M tokens | $2.50 | $10 |
| Results tracked | 54 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
Gemini 2.5 Flash: 35.8 (#220), GPT-5: 50.3 (#47)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| SWE-bench Verified (bash only) | 28.7% | 65% |
| Aider Polyglot | 55.1% | 88% |
| WeirdML | 41.9% | 60.7% |
| LMArena Coding | 1424 | 1436 |
| ALE-Bench | 661.88 | 1,162 |
| SWE-bench Verified | — | 73.6% |
| LMArena WebDev | — | 1418 |
| SciCode | — | 42.9% |
| GSO | — | 6.9% |
| AlgoTune | — | 1.67 |
Agentic & Tool Use GPT-5 leads
Gemini 2.5 Flash: 30.8 (#74), GPT-5: 33.1 (#56)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| Terminal-Bench | 17.1% | 49.6% |
| BALROG | 33.5% | 32.8% |
| Berkeley Function Calling Leaderboard | 56.2% | — |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| TheAgentCompany | 41.1% | — |
| DeepResearch Bench | — | 49.6% |
| LMArena Search | — | 1133 |
| METR Time Horizons | — | 69.6% |
| Vending-Bench 2 | 548.84 | — |
Reasoning GPT-5 leads
Gemini 2.5 Flash: 18.1 (#286), GPT-5: 38.3 (#64)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| ARC-AGI-2 | 2.5% | 9.9% |
| SimpleBench | 41.2% | 56.7% |
| Kagi LLM Benchmark | 56.8% | 72.7% |
| ARC-AGI-1 | 33.3% | 65.7% |
| CritPt | 1.1% | 12.6% |
| EnigmaEval | 2.7% | 10.5% |
| LMArena Hard Prompts | 1422 | 1416 |
| DTBench | 76.5% | 90.7% |
| LMCA | 27.5% | 40% |
| Epoch Capabilities Index | 143.03 | 150 |
| ForecastBench | 60.6 | 61.4 |
| Chess Puzzles | — | 37% |
| EBR-Bench | — | 12.7% |
| Mystery Game Puzzles | — | 23% |
Math GPT-5 leads
Gemini 2.5 Flash: 39.9 (#98), GPT-5: 55.0 (#44)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.1% | 91.4% |
| Omni-MATH | 38.5% | 64.7% |
| LMArena Math | 1415 | 1407 |
| FrontierMath (Feb 2025 set) | 4.8% | 32.4% |
| FrontierMath Tier 4 (v1) | 4.2% | 12.5% |
| FrontierMath (Tiers 1-3) | — | 55.4% |
| FrontierMath Tier 4 | — | 22% |
| ProofBench | — | 18% |
| MATH Level 5 | — | 98.1% |
Knowledge GPT-5 leads
Gemini 2.5 Flash: 36.4 (#168), GPT-5: 56.6 (#43)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| Humanity's Last Exam | 12.1% | 25.3% |
| MMLU-Pro | 63.9% | 86.3% |
| Confabulations | 16.8% | 10.3% |
| Vectara Hallucination Rate | 7.8% | 14.7% |
| GPQA (HELM) | 39% | 79.2% |
| LMArena Expert | 1426 | 1419 |
| GPQA Diamond | — | 86.2% |
| SimpleQA Verified | — | 50.1% |
Multimodal GPT-5 leads
Gemini 2.5 Flash: 41.8 (#32), GPT-5: 46.8 (#13)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| LMArena Vision | 1253 | 1232 |
| GeoBench | 76% | 81% |
| VPCT | 46.2% | 66% |
| SpatialViz-Bench | 36.9% | — |
Multilingual Too close to call
Gemini 2.5 Flash: 52.3 (#88), GPT-5: 51.4 (#110)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| LMArena Non-English | 1409 | 1397 |
| LMArena Chinese | 1450 | 1422 |
| LMArena French | 1433 | 1410 |
| LMArena German | 1418 | 1416 |
| LMArena Japanese | 1405 | 1409 |
| LMArena Korean | 1385 | 1360 |
| LMArena Russian | 1415 | 1406 |
| LMArena Spanish | 1421 | 1399 |
Instruction Following Gemini 2.5 Flash leads
Gemini 2.5 Flash: 75.7 (#54), GPT-5: 73.8 (#113)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| IFEval | 89.8% | 87.5% |
| LMArena Instruction Following | 1405 | 1388 |
Long Context GPT-5 leads
Gemini 2.5 Flash: 47.5 (#17), GPT-5: 69.5 (#2)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| Fiction.LiveBench | 77.8% | 97.2% |
| LMArena Longer Query | 1419 | 1399 |
Writing & Preference GPT-5 leads
Gemini 2.5 Flash: 53.8 (#157), GPT-5: 63.4 (#65)
| Benchmark | Gemini 2.5 Flash | GPT-5 |
|---|---|---|
| LMArena Text | 1417 | 1406 |
| LMArena Creative Writing | 1400 | 1365 |
| Short-Story Creative Writing | 76.5% | 86% |
| EQ-Bench Creative Writing | 1137 | 1627 |
| WildBench | 81.7% | 85.7% |
| LMArena Multi-Turn | 1408 | 1426 |
Frequently asked questions
Is Gemini 2.5 Flash better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 39.3 on the Noometry Index. Gemini 2.5 Flash costs 4.0× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, Gemini 2.5 Flash or GPT-5?
Gemini 2.5 Flash is cheaper. It lists at $0.30 per million input tokens and $2.50 per million output tokens; GPT-5 lists at $1.25 and $10.
Is Gemini 2.5 Flash or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 35.8 in the Noometry coding category.
Which has the bigger context window?
Gemini 2.5 Flash does, with 1.05M tokens against 400K.
How many benchmarks do Gemini 2.5 Flash and GPT-5 share?
50 benchmarks have published results for both models. Gemini 2.5 Flash has 54 scored results on Noometry and GPT-5 has 69.