OpenAI model product
GPT‑5.2 introduces substantial gains in professional knowledge work, outperforming experts on GDPval with 70.9% wins or ties, and setting new highs in coding (SWE‑Bench Pro 55.6%), science (GPQA Diamond ~92–93%), math (AIME 2025: 100%), long‑context accuracy up to 256k tokens, and reliable tool‑calling (Tau2 Telecom 98.7%). It rolls out as Instant, Thinking, and Pro—faster, more structured, and less error‑prone—priced at $1.75/1M input and $14/1M output tokens, with Pro variants supporting xhigh reasoning for top‑quality, end‑to‑end execution.
Updated Aug 10, 2026. Default version: GPT-5.2
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version GPT-5.2.
| BrowseComp Long Context 128k | 0.9 | 1 | 5 | 100.0% | C | |
| BrowseComp Long Context 256k | 0.9 | 1 | 2 | 100.0% | C | |
| Graphwalks BFS <128k | 0.9 | 1 | 22 | 100.0% | C | |
| AIME 2025 | 1.0 | 2 | 114 | 99.1% | C | |
| HMMT 2025 | 1.0 | 2 | 33 | 96.9% | C | |
| ScreenSpot Pro | 0.9 | 2 | 24 | 95.7% | C | |
| Graphwalks parents <128k | 0.9 | 3 | 18 | 88.2% | C | |
| SWE-Lancer (IC-Diamond subset) | 0.7 | 3 | 6 | 60.0% | C | |
| Tau2 Telecom | 1.0 | 4 | 35 | 91.2% | C | |
| VideoMMMU | 0.9 | 4 | 26 | 88.0% | C | |
| ARC-AGI | 0.9 | 5 | 7 | 33.3% | C | |
| FrontierMath | 0.4 | 5 | 17 | 75.0% | C | |
| Tau2 Retail | 0.8 | 6 | 26 | 80.0% | C | |
| ARC-AGI v2 | 0.5 | 8 | 16 | 53.3% | C | |
| MMMLU | 0.9 | 9 | 49 | 83.3% | C | |
| GPQA | 0.9 | 13 | 233 | 94.8% | C | |
| CharXiv-R | 0.8 | 14 | 47 | 71.7% | C | |
| SWE-Bench Verified | 0.8 | 14 | 104 | 87.4% | C | |
| MMMU-Pro | 0.8 | 15 | 65 | 78.1% | C | |
| LiveBench | 0.7 | 16 | 38 | 59.5% | B | |
| Toolathlon | 0.5 | 20 | 31 | 36.7% | C | |
| MCP Atlas | 0.6 | 26 | 30 | 13.8% | C | |
| BrowseComp | 0.7 | 31 | 58 | 47.4% | C | |
| Humanity's Last Exam | 0.3 | 41 | 92 | 56.0% | C |
Preference and agent-evaluation signals from published Arena datasets.
The default version has no published Arena rows, or its source alias has not been resolved.
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
| OpenAI | gpt-5.2 | global | $1.75 | $14 | 400K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| GPT-5.2 | 69.1 | N/A | 400K | 128K | No | Proprietary |
A concise description based on the published model registry.
GPT‑5.2 introduces substantial gains in professional knowledge work, outperforming experts on GDPval with 70.9% wins or ties, and setting new highs in coding (SWE‑Bench Pro 55.6%), science (GPQA Diamond ~92–93%), math (AIME 2025: 100%), long‑context accuracy up to 256k tokens, and reliable tool‑calling (Tau2 Telecom 98.7%). It rolls out as Instant, Thinking, and Pro—faster, more structured, and less error‑prone—priced at $1.75/1M input and $14/1M output tokens, with Pro variants supporting xhigh reasoning for top‑quality, end‑to‑end execution.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about GPT-5.2.
GPT-5.2's default version was released on Dec 11, 2025.
GPT-5.2's official API price is $1.75 per million input tokens and $14 per million output tokens via OpenAI.
GPT-5.2 is published under OpenAI in the model registry.
The default version has a 400K token context window.
No. The default version is not marked as having publicly available weights.
1 published provider offerings are linked to the default version.
Nearby ranked alternatives include Llama 3.1 405B, DeepSeek-V2.5, Claude Opus 4.6.