Model comparison
Gemini 2.5 Pro vs o3
o3 is the stronger model overall, scoring 47.5 to 45.0 on the Noometry Index.
Last verified . 58 shared benchmarks.
GoogleSummary
- They share 58 benchmarks with published results for both. Gemini 2.5 Pro scores higher in 6 categories and o3 in 4 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o3 leads 50.2 to 32.5.
- The biggest single-benchmark swing is Omni-MATH: 41.6% for Gemini 2.5 Pro and 71.4% for o3.
- Both cost about the same: $1.25 input and $10 output per million tokens.
- Gemini 2.5 Pro accepts more context: 1.05M tokens versus 200K.
Side by side
| Gemini 2.5 Pro | o3 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 45.0 | 47.5 |
| Released | 2025-03-25 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 200K |
| Max output | 66K | 100K |
| Input $ / M tokens | $1.25 | $2 |
| Output $ / M tokens | $10 | $8 |
| Results tracked | 78 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
Gemini 2.5 Pro: 42.4 (#101), o3: 46.8 (#64)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| SWE-bench Verified | 57.6% | 62.3% |
| SWE-bench Verified (bash only) | 53.6% | 58.4% |
| Aider Polyglot | 83.1% | 81.3% |
| GSO | 3.9% | 8.8% |
| WeirdML | 54% | 52.4% |
| LMArena Coding | 1452 | 1408 |
| CadEval | 64% | 74% |
| ALE-Bench | 785.52 | 933.55 |
| LMArena WebDev | 1227 | — |
| SciCode | 42.8% | — |
| LiveBench Coding | 85.9% | — |
| AlgoTune | 1.51 | — |
Agentic & Tool Use o3 leads
Gemini 2.5 Pro: 29.2 (#88), o3: 34.5 (#44)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| GDPval | 23.3% | 30.8% |
| DeepResearch Bench | 42.8% | 45.2% |
| LMArena Search | 1142 | 1144 |
| METR Time Horizons | 55.4% | 65.4% |
| Terminal-Bench | 32.6% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| Remote Labor Index | 0.8% | — |
| TheAgentCompany | 30.3% | — |
| τ²-bench Banking | 13.7% | — |
| OSWorld | — | 23% |
| BALROG | 43.3% | — |
| Vending-Bench 2 | 573.64 | — |
Reasoning o3 leads
Gemini 2.5 Pro: 28.8 (#99), o3: 32.0 (#78)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| ARC-AGI-2 | 4.9% | 6.5% |
| SimpleBench | 62.4% | 53.1% |
| Kagi LLM Benchmark | 70.3% | 67.6% |
| ARC-AGI-1 | 41% | 60.8% |
| CritPt | 2% | 1.4% |
| Chess Puzzles | 20% | 38% |
| EnigmaEval | 5.6% | 13.1% |
| LMArena Hard Prompts | 1455 | 1402 |
| DTBench | 82.4% | 84.8% |
| LMCA | 34.8% | 39.7% |
| Epoch Capabilities Index | 145.32 | 146.86 |
| ForecastBench | 61.3 | 62.5 |
| LiveBench Reasoning | 89.8% | — |
| Mystery Game Puzzles | — | 29% |
| LiveBench Data Analysis | 79.9% | — |
| LiveBench | 82.3% | — |
Math o3 leads
Gemini 2.5 Pro: 32.5 (#213), o3: 50.2 (#58)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 24.6% | 33.3% |
| OTIS Mock AIME 2024-2025 | 84.7% | 84.4% |
| Omni-MATH | 41.6% | 71.4% |
| LMArena Math | 1450 | 1426 |
| MATH Level 5 | 95.9% | 97.8% |
| FrontierMath (Feb 2025 set) | 14.1% | 18.7% |
| FrontierMath Tier 4 (v1) | 4.2% | 2.1% |
| FrontierMath Tier 4 | 0% | — |
| LiveBench Math | 90.2% | — |
Knowledge Gemini 2.5 Pro leads
Gemini 2.5 Pro: 56.0 (#46), o3: 54.6 (#52)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| GPQA Diamond | 85.3% | 81.8% |
| Humanity's Last Exam | 21.6% | 20.3% |
| MMLU-Pro | 86.3% | 85.9% |
| Confabulations | 10.6% | 14.4% |
| GPQA (HELM) | 74.9% | 75.3% |
| LMArena Expert | 1452 | 1402 |
| SimpleQA Verified | — | 49.4% |
| Vectara Hallucination Rate | 7% | — |
Multimodal Gemini 2.5 Pro leads
Gemini 2.5 Pro: 45.2 (#18), o3: 41.4 (#36)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| LMArena Vision | 1263 | 1214 |
| GeoBench | 86% | 74% |
| VPCT | 48% | 52% |
| LMArena Document | 1421 | — |
| SpatialViz-Bench | 44.7% | — |
Multilingual Gemini 2.5 Pro leads
Gemini 2.5 Pro: 55.3 (#31), o3: 51.7 (#105)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| LMArena Non-English | 1451 | 1401 |
| LMArena Chinese | 1507 | 1437 |
| LMArena French | 1472 | 1430 |
| LMArena German | 1487 | 1420 |
| LMArena Japanese | 1461 | 1403 |
| LMArena Korean | 1434 | 1370 |
| LMArena Russian | 1461 | 1406 |
| LMArena Spanish | 1473 | 1395 |
Instruction Following Gemini 2.5 Pro leads
Gemini 2.5 Pro: 75.0 (#75), o3: 72.8 (#127)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| IFEval | 84% | 86.9% |
| LMArena Instruction Following | 1437 | 1368 |
| LiveBench Instruction Following | 80.6% | — |
Long Context Gemini 2.5 Pro leads
Gemini 2.5 Pro: 59.8 (#5), o3: 53.3 (#6)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| Fiction.LiveBench | 91.7% | 88.9% |
| LMArena Longer Query | 1449 | 1372 |
| CL-bench | — | 17.8% |
Writing & Preference Too close to call
Gemini 2.5 Pro: 63.7 (#62), o3: 63.5 (#64)
| Benchmark | Gemini 2.5 Pro | o3 |
|---|---|---|
| LMArena Text | 1458 | 1410 |
| LMArena Creative Writing | 1454 | 1359 |
| Short-Story Creative Writing | 83.8% | 83.9% |
| EQ-Bench Creative Writing | 1421 | 1676 |
| WildBench | 85.7% | 86.1% |
| LMArena Multi-Turn | 1453 | 1405 |
| LiveBench Language | 67.8% | — |
Frequently asked questions
Is Gemini 2.5 Pro better than o3?
o3 is the stronger model overall, scoring 47.5 to 45.0 on the Noometry Index.
Which is cheaper, Gemini 2.5 Pro or o3?
Gemini 2.5 Pro is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; o3 lists at $2 and $8.
Is Gemini 2.5 Pro or o3 better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Gemini 2.5 Pro does, with 1.05M tokens against 200K.
How many benchmarks do Gemini 2.5 Pro and o3 share?
58 benchmarks have published results for both models. Gemini 2.5 Pro has 78 scored results on Noometry and o3 has 63.