GPT-5.2vsGPT-5.1
Across 13 shared benchmarks, GPT-5.2 leads overall: GPT-5.2 wins 13, GPT-5.1 wins 0, with 0 ties and an average score difference of +10.48.
GPT-5.213 wins(100%)(0%)0 winsGPT-5.1
Benchmark scores
Grouped by capability, sorted by largest gap within each. 13 shared benchmarks.
General Knowledge
GPT-5.2 4/4| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| ARC-AGI-2 | 54.2020 / 59深度思考（无工具、并行） | 17.6033 / 59 | +36.60 |
| HLE | 45.5032 / 157Deep Thinking (With Tools + Internet) | 26.5083 / 157 | +19 |
| ARC-AGI | 90.5015 / 65深度思考（无工具、并行） | 72.8025 / 65 | +17.70 |
| GPQA Diamond | 93.208 / 178深度思考（无工具、并行） | 88.1028 / 178 | +5.10 |
Coding and Software Engineer
GPT-5.2 3/3| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| IC SWE-Lancer(Diamond) | 74.602 / 8极高强度思考（工具） | 69.703 / 8Thinking High (No Tools) | +4.90 |
| SWE-Bench Pro - Public | 55.6017 / 43极高强度思考（工具） | 50.8030 / 43Thinking High (No Tools) | +4.80 |
| SWE-bench Verified | 8016 / 108极高强度思考（工具） | 76.3030 / 108 | +3.70 |
Math and Reasoning
GPT-5.2 3/3| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| FrontierMath | 40.308 / 60极高强度思考（工具） | 26.7013 / 60Thinking High (With Tools) | +13.60 |
| FrontierMath - Tier 4 | 18.8016 / 80Thinking High (No Tools) | 12.5029 / 80Thinking High (With Tools) | +6.30 |
| AIME2025 | 1001 / 106极高强度思考（无工具） | 9428 / 106 | +6 |
Agent Level Benchmark
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 98.704 / 35极高强度思考（工具） | 95.6014 / 35Thinking High (With Tools) | +3.10 |
AI Agent - Information Search
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| BrowseComp | 65.8024 / 45Deep Thinking (With Tools + Internet) | 50.8036 / 45Thinking High (No Tools) | +15 |
Multimodal Understanding
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| MMMU | 85.901 / 28极高强度思考（无工具） | 85.402 / 28 | +0.50 |
Specs
| Field | GPT-5.2 | GPT-5.1 |
|---|---|---|
| Publisher | OpenAI | OpenAI |
| Release date | 2025-12-11 | 2025-11-12 |
| Model type | Chat model | Reasoning model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 400K | 400K |
| Max output | Not available | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | GPT-5.2 | GPT-5.1 |
|---|---|---|
| Text input | $1.75 / 1M tokens | Not public |
| Text output | $14 / 1M tokens | Not public |
| Cache read | $0.175 / 1M tokens | Not public |
| Cache write | $1.75 / 1M tokens | Not public |
One or both models have incomplete public pricing.
Summary
- GPT-5.2leads in:General Knowledge (4/4), Coding and Software Engineer (3/3), Math and Reasoning (3/3), Agent Level Benchmark (1/1), AI Agent - Information Search (1/1), Multimodal Understanding (1/1)
On average across the 13 shared benchmarks, GPT-5.2 scores 10.48 higher.
Largest single-benchmark gap: ARC-AGI-2 — GPT-5.2 54.20 vs GPT-5.1 17.60 (+36.60).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.