GPT-5.2vsGPT-5.1
Across 4 shared benchmarks, GPT-5.2 leads overall: GPT-5.2 wins 3, GPT-5.1 wins 1, with 0 ties and an average score difference of +4.71.
GPT-5.23 wins(75%)(25%)1 winGPT-5.1
Benchmark scores
Grouped by capability, sorted by largest gap within each. 4 shared benchmarks.
Coding and Software Engineer
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| GSO | 27.407 / 21Thinking High (With Tools) | 13.7013 / 21Thinking High (With Tools) | +13.70 |
Commonsense Reasoning
GPT-5.1 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| SimpleBench | 45.8058 / 93Thinking High (No Tools) | 53.2045 / 93Thinking High (No Tools) | -7.40 |
General Knowledge
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| LiveBench | 48.9196 / 117Normal (No Tools) | 42.65108 / 117Normal (No Tools) | +6.26 |
Math and Reasoning
GPT-5.2 1/1| Benchmark | GPT-5.2 | GPT-5.1 | Diff |
|---|---|---|---|
| FrontierMath - Tier 4 | 18.8016 / 80Thinking High (No Tools) | 12.5029 / 80Thinking High (No Tools) | +6.30 |
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 | $1.25 / 1M tokens |
| Text output | $14 / 1M tokens | $10 / 1M tokens |
| Cache read | $0.175 / 1M tokens | $0.125 / 1M tokens |
| Cache write | $1.75 / 1M tokens | $0 / 1M tokens |
Summary
- GPT-5.2leads in:Coding and Software Engineer (1/1), General Knowledge (1/1), Math and Reasoning (1/1)
- GPT-5.1leads in:Commonsense Reasoning (1/1)
On average across the 4 shared benchmarks, GPT-5.2 scores 4.71 higher.
Largest single-benchmark gap: GSO — GPT-5.2 27.40 vs GPT-5.1 13.70 (+13.70).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.