Model comparison
Gemini 3.1 Pro vs GPT-5.2
Head-to-head evidence from 19 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: Gemini 3.1 Pro #83 (Estimated); GPT-5.2 #64 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. Gemini 3.1 Pro and GPT-5.2 share 19 comparable benchmark results. 3 of 8 categories are comparable. 28 results are unique to Gemini 3.1 Pro; 9 to GPT-5.2.
Updated July 20, 2026- Shared results
- 19
- Gemini 3.1 Pro only
- 28
- GPT-5.2 only
- 9
- Comparable categories
- 3 / 8
Pick GPT-5.2 if you want the stronger benchmark profile. Gemini 3.1 Pro only becomes the better choice if reasoning is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 19 shared benchmark results across 7 evidence categories; 3 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GPT-5.2 is clearly ahead on the BenchAlign aggregate, 58.43 to 55.3. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
GPT-5.2's sharpest advantage is in mathematics, where it averages 35.2 against 31.8. The single biggest benchmark swing on the page is ARC-AGI-2, 77.1% to 52.9%. Gemini 3.1 Pro does hit back in reasoning, so the answer changes if that is the part of the workload you care about most.
GPT-5.2 is also the more expensive model on tokens at $1.75 input / $14.00 output per 1M tokens, versus $2.00 input / $12.00 output per 1M tokens for Gemini 3.1 Pro. GPT-5.2 is the reasoning model in the pair, while Gemini 3.1 Pro is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. Gemini 3.1 Pro gives you the larger context window at 1M, compared with 400K for GPT-5.2.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | Gemini 3.1 Pro | Δ | GPT-5.2 |
|---|---|---|---|
| Reasoning | Gemini 3.1 Pro77.1 | Margin← 24.2 | GPT-5.252.9 |
| Math | Gemini 3.1 Pro31.8 | Margin→ 3.4 | GPT-5.235.2 |
| Multimodal | Gemini 3.1 Pro82.6 | Margin← 2.2 | GPT-5.280.4 |
| Agentic | Gemini 3.1 ProNot measured | MarginNo overlap | GPT-5.255.7 |
| Coding | Gemini 3.1 ProNot measured | MarginNo overlap | GPT-5.270.6 |
| Knowledge | Gemini 3.1 ProNot measured | MarginNo overlap | GPT-5.292.4 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
ARC-AGI-2
ReasoningA 77.1%B 52.9%Winner: Gemini 3.1 ProΔ 24.2ARC-AGI-2: Gemini 3.1 Pro scored 77.1%; GPT-5.2 scored 52.9%. Gemini 3.1 Pro wins this benchmark. - Source ↗
MMMU-Pro
MultimodalA 83.9%B 79.5%Winner: Gemini 3.1 ProΔ 4.4MMMU-Pro: Gemini 3.1 Pro scored 83.9%; GPT-5.2 scored 79.5%. Gemini 3.1 Pro wins this benchmark. - Source ↗
FrontierMath v2 (Tiers 1-3)
MathA 36.900%B 40.700%Winner: GPT-5.2Δ 3.8FrontierMath v2 (Tiers 1-3): Gemini 3.1 Pro scored 36.900%; GPT-5.2 scored 40.700%. GPT-5.2 wins this benchmark. - Source ↗
FrontierMath v2 (Tier 4)
MathA 16.700%B 18.800%Winner: GPT-5.2Δ 2.1FrontierMath v2 (Tier 4): Gemini 3.1 Pro scored 16.700%; GPT-5.2 scored 18.800%. GPT-5.2 wins this benchmark. - Source ↗
CharXiv
MultimodalA 80.2%B 82.1%Winner: GPT-5.2Δ 1.9CharXiv: Gemini 3.1 Pro scored 80.2%; GPT-5.2 scored 82.1%. GPT-5.2 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | Gemini 3.1 Pro | GPT-5.2 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | Gemini 3.1 Pro$2 input / $12 output | GPT-5.2$1.75 input / $14 output | Gemini 3.1 Pro has the lower combined listed price. |
| Generation speedtokens per second | Gemini 3.1 Pro109 tok/s | GPT-5.273 tok/s | Gemini 3.1 Pro has the higher measured throughput. |
| First-answer latencyseconds to first token | Gemini 3.1 Pro29.71 s | GPT-5.2130.34 s | Gemini 3.1 Pro reaches the first token sooner. |
| Context windowmaximum listed tokens | Gemini 3.1 Pro1M | GPT-5.2400K | Gemini 3.1 Pro lists the larger context window. |
Benchmark Deep Dive
Agentic19 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| Claw-EvalSource | 57.8% | — | Not comparable |
| DeepSearchQASource | 69.7% | — | Not comparable |
| τ²-bench resultsSource | 95.6% | 84.8% | Gemini 3.1 Pro leads |
| AA Agentic IndexSource | 21.4% | — | Not comparable |
| APEX-Agents-AASource | 32.0% | — | Not comparable |
| GDPval-AASource | 23.2% | — | Not comparable |
| GDPval-AASource | 965 | — | Not comparable |
| Gert LabsSource | 56.87% | 46.54% | Gemini 3.1 Pro leads |
| ResearchClawBenchSource | 13.3% | — | Not comparable |
| AA AutomationBenchSource | 37.5% | — | Not comparable |
| AA EnterpriseOps-GymSource | 42.2% | — | Not comparable |
| AA Harvey LABSource | 0.0% | — | Not comparable |
| AA ITBenchSource | 30.3% | — | Not comparable |
| AA Tau3 BankingSource | 16.5% | — | Not comparable |
| terminalBenchHardSource | 53.8% | — | Not comparable |
| aaTerminalBench21Source | 73.8% | — | Not comparable |
| BrowseCompSource | — | 65.8% | Not comparable |
| OSWorld-VerifiedSource | — | 47.3% | Not comparable |
| JobBenchSource | — | 34.3% | Not comparable |
Coding7 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| LiveCodeBench ProSource | 82.9% | — | Not comparable |
| React Native EvalsSource | 78.9% | — | Not comparable |
| Vibe Code BenchSource | 32.03% | 53.50% | GPT-5.2 leads |
| AA Coding IndexSource | 68.8% | — | Not comparable |
| AA-SciCodeSource | 58.9% | 52.1% | Gemini 3.1 Pro leads |
| SWE-bench VerifiedSource | — | 80% | Not comparable |
| SWE-bench ProSource | — | 55.6% | Not comparable |
ReasoningGemini 3.1 Pro wins3 benchmarks
Knowledge11 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| GPQA-DSource | 94.3% | — | Not comparable |
| HLE w/o toolsSource | 45.4% | — | Not comparable |
| HealthBench HardSource | 20.6% | — | Not comparable |
| MedXpertQA (Text)Source | 71.5% | — | Not comparable |
| Artificial Analysis Intelligence IndexSource | 46.5% | 42.2% | Gemini 3.1 Pro leads |
| AA-GPQA DiamondSource | 94.1% | 90.3% | Gemini 3.1 Pro leads |
| AA-HLESource | 44.7% | 35.4% | Gemini 3.1 Pro leads |
| AA-Omniscience IndexSource | 32.9% | -1.0% | Gemini 3.1 Pro leads |
| AA-Omniscience AccuracySource | 55.3% | 43.8% | Gemini 3.1 Pro leads |
| AA-Omniscience Hallucination RateSource | 49.9% | 79.7% | Gemini 3.1 Pro leads |
| GPQASource | — | 92.4% | Not comparable |
MathGPT-5.2 wins3 benchmarks
Multilingual1 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| AA Global-MMLU-LiteSource | 93.2% | — | Not comparable |
MultimodalGemini 3.1 Pro wins11 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| MMMU-ProSource | 83.9% | 79.5% | Gemini 3.1 Pro leads |
| CharXivSource | 80.2% | 82.1% | GPT-5.2 leads |
| ERQASource | 69.4% | — | Not comparable |
| SimpleVQASource | 72.4% | — | Not comparable |
| ScreenSpot ProSource | 84.4% | — | Not comparable |
| ZeroBenchSource | 29.0% | — | Not comparable |
| MedXpertQA (MM)Source | 81.3% | — | Not comparable |
| AA-MMMU-ProSource | 82.4% | — | Not comparable |
| Design Arena WebsiteSource | 1284 | 1227 | Gemini 3.1 Pro leads |
| MathVisionSource | — | 83.0% | Not comparable |
| V*Source | — | 75.9% | Not comparable |
Inst. Following1 benchmarks
| Benchmark | Gemini 3.1 Pro | GPT-5.2 | Result |
|---|---|---|---|
| AA-IFBenchSource | 77.1% | 75.4% | Gemini 3.1 Pro leads |
Frequently Asked Questions (4)
Which is better, Gemini 3.1 Pro or GPT-5.2?
GPT-5.2 is ahead on BenchLM's BenchAlign leaderboard, 58.43 to 55.3. The biggest single separator in this matchup is ARC-AGI-2, where the scores are 77.1% and 52.9%.
Which is better for math, Gemini 3.1 Pro or GPT-5.2?
GPT-5.2 has the edge for math in this comparison, averaging 35.2 versus 31.8. Inside this category, FrontierMath v2 (Tiers 1-3) is the benchmark that creates the most daylight between them.
Which is better for reasoning, Gemini 3.1 Pro or GPT-5.2?
Gemini 3.1 Pro has the edge for reasoning in this comparison, averaging 77.1 versus 52.9. Inside this category, ARC-AGI-2 is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, Gemini 3.1 Pro or GPT-5.2?
Gemini 3.1 Pro has the edge for multimodal and grounded tasks in this comparison, averaging 82.6 versus 80.4. Inside this category, Design Arena Website is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.