GPT-5.2
GPT-5.2 is an AI model from OpenAI.
- Publisher
- OpenAI
- Family
- GPT
- Context
- 400K
- Input
- Output
- License
- proprietary
- Released
- 11 Dec 2025
Cite
Notes
Only stored in your browser.
Attribution
- Benchmark scores
- prime-hubfrontiermathAALiveBenchSWE-benchmatharenaOpenRewardAfterQueryVals AIEQ-BenchLech Mazur
Context
400K
Weights
Closed
API only
Input
$1.75
/1M tok
Output
$14
/1M tok
Cache read
$0.17
0.10x input
Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter
| Reasoning effort | Index | Speed |
|---|---|---|
| xhigh | 30.4 | 0 t/s |
| medium | 26.5 | 0 t/s |
| Non-reasoning | 17.0 | 0 t/s |
Intelligence Index per reasoning-effort setting, via Artificial Analysis · pricing identical across tiers
Reported on 43 evals across 13 domains - best 100.0% on Bb Demo
Reported eval scores
43AIME2025100
Hanabi12.9
Curvebench Env40.6
LiveCodeBench88.9
MMLU-Pro87.4
LiveBench78.7
LiveBench - Math93.2
IFBench75.4
Head-to-heads
29benchmark comparisons against other models
vs Claude Opus 4.535 shared evalsvs Claude Sonnet 4.532 shared evalsvs GPT-5 Mini32 shared evalsvs Claude Opus 4.630 shared evalsvs Grok 4.1 Fast30 shared evalsvs Gemini 3.1 Pro Preview29 shared evalsvs GPT-5.429 shared evalsvs Kimi K2.529 shared evalsvs DeepSeek V3.229 shared evalsvs GPT-5.527 shared evalsvs GPT-5 Nano26 shared evalsvs Claude Opus 4.826 shared evalsvs Claude Opus 4.726 shared evalsvs Claude 4 Sonnet25 shared evalsvs GLM 5.125 shared evalsvs GPT-5.4 Mini25 shared evalsvs Claude Fable 524 shared evalsvs Claude Sonnet 4.624 shared evalsvs Gemini 2.5 Pro24 shared evalsvs GLM 4.724 shared evalsvs GPT-523 shared evalsvs Qwen3.6 Plus23 shared evalsvs gpt-oss-120b22 shared evalsvs GPT-4o18 shared evalsvs GPT-4.1 Mini17 shared evalsvs GPT-4.115 shared evalsvs GPT-4o-mini15 shared evalsvs GPT-4.1 Nano15 shared evalsvs GPT-6 Astra13 shared evals