Head to head
GPT-4.1 vs Gemini 2.5 Pro
GPT-4.1 from OpenAI against Gemini 2.5 Pro from Google DeepMind — specification, price and every benchmark both makers have published, in one table.
Benchmarks
Gemini 2.5 Pro leads
Gemini 2.5 Pro wins 3 of the 3 benchmarks both models report, GPT-4.1 wins 0, by a wide margin. The average gap across shared tests is 11.3 points.
Price
Gemini 2.5 Pro is cheaper
On a 3:1 input-to-output mix, Gemini 2.5 Pro costs $3.44 per million tokens against $3.50 for GPT-4.1 — a small difference. Remember that a reasoning model bills its thinking as output, so cost per answer can diverge much further than cost per token.
What actually differs
- Gemini 2.5 Pro takes 1M tokens of context against 1M — 1.0× more room for long documents or a large codebase.
- Gemini 2.5 Pro is a reasoning model and the other is not, which usually means better maths and multi-step logic in exchange for higher latency and more billed output tokens.
Scorecard
Which is better at what
Maths, coding, reasoning and the rest — one line each, averaged over the benchmarks both models actually report.
| Category | GPT-4.1 | Gemini 2.5 Pro | Better at this |
|---|---|---|---|
| ReasoningGPQA Diamond | 66.3% | 84.0% | Gemini 2.5 Pro+17.7 |
| MathsOnly one model reports this | — | — | Not comparable |
| CodingSWE-bench Verified | 54.6% | 63.8% | Gemini 2.5 Pro+9.2 |
| KnowledgeOnly one model reports this | — | — | Not comparable |
| MultimodalMMMU | 74.8% | 81.7% | Gemini 2.5 Pro+6.9 |
| Instruction followingOnly one model reports this | — | — | Not comparable |
| Human preferenceOnly one model reports this | — | — | Not comparable |
| Categories wonOut of 3 comparable | 0 | 3 | Gemini 2.5 Pro |
Each category averages only the benchmarks every model here reports, so no one is credited for a test the other did not run. A category with no shared test is marked Not comparable rather than guessed at.
Side by side
GPT-4.1 and Gemini 2.5 Pro, row by row
| Attribute | GPT-4.1OpenAI | Gemini 2.5 ProGoogle DeepMind |
|---|---|---|
| Specification | ||
| MakerWho built it | OpenAI | Google DeepMind |
| Released | 2025-04 | 2025-03 |
| Context windowHow much can go in at once | 1,047,576 tokens | 1,048,576 tokens |
| Max output | 32,768 tokens | 65,536 tokens |
| Input | Text, Image | Text, Image, Audio, Video |
| ReasoningSpends extra tokens thinking before it answers | No | Yes |
| Tool calling | Yes | Yes |
| Knowledge cutoff | 2024-06 | 2025-01 |
| Licence | Proprietary | Proprietary |
| Open weightsCan you download and run it yourself | No | No |
| Price | ||
| Input priceUSD per million tokens in | $2.00 | $1.25 |
| Output priceUSD per million tokens out | $8.00 | $10.00 |
| Cached input | $0.50 | $0.31 |
| Blended 3:1A 3-in-to-1-out million-token mix — a fairer single number than input price alone | $3.50 | $3.44Cheapest |
| Published benchmarks | ||
| MMLU-Pro12,000 reasoning-heavy multiple-choice questions across 14 academic subjects, with ten options instead of four. The harder successor to MMLU. | Not reported | |
| GPQA Diamond198 graduate-level physics, chemistry and biology questions written to be Google-proof. PhD holders in the matching field score about 65%. | ||
| AIME 2025The American Invitational Mathematics Examination — 15 problems, integer answers, no partial credit. A standard test of multi-step maths reasoning. | Not reported | |
| SWE-bench Verified500 human-validated GitHub issues from real Python repositories. The model must produce a patch that makes the project's own tests pass. | ||
| MMMUCollege-level questions that require reading charts, diagrams, tables and photographs alongside the text. | ||
| IFEvalVerifiable instructions — word counts, formats, forbidden words — checked by a program rather than a judge model. | Not reported | |
| LMArena EloElo rating from blind pairwise votes by the public on LMArena. Measures what people prefer, not what is correct. | Not reported | |
| Links | Full page | Full page |
| Row verified | 2026-08 | 2026-08 |
Where these numbers come from
Every score on this page is a published figure, taken from the model's own card, system card, technical report or release post, or from a public leaderboard. CorX Labs did not run these evaluations. Most are self-reported by the lab that built the model, which means they were produced under that lab's own choice of prompt, scaffold and number of attempts — so treat them as a starting point for a shortlist, not as a settled ranking.
A score someone other than the model's maker measured is marked Independent and names its measurer. Those are the stronger numbers on this page — an outside harness has no reason to flatter anyone — and there are not many of them.
Where a figure has not been published, the cell reads Not reported rather than an estimate. Nothing here is inferred, interpolated or guessed. Each model records the month its row was last checked. Full method and caveats.
Questions
GPT-4.1 or Gemini 2.5 Pro?
Is GPT-4.1 better than Gemini 2.5 Pro?
Gemini 2.5 Pro wins 3 of the 3 benchmarks both models report, GPT-4.1 wins 0, by a wide margin. The average gap across shared tests is 11.3 points.
Which is cheaper, GPT-4.1 or Gemini 2.5 Pro?
On a 3:1 input-to-output mix, Gemini 2.5 Pro costs $3.44 per million tokens against $3.50 for GPT-4.1 — a small difference. Remember that a reasoning model bills its thinking as output, so cost per answer can diverge much further than cost per token.
What is the difference between GPT-4.1 and Gemini 2.5 Pro?
Gemini 2.5 Pro takes 1M tokens of context against 1M — 1.0× more room for long documents or a large codebase. Gemini 2.5 Pro is a reasoning model and the other is not, which usually means better maths and multi-step logic in exchange for higher latency and more billed output tokens.
