GPT-5.2 (Xhigh)
OpenAIGPTProprietary
Description
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.
Date de sortie
2025-12-11
Paramètres
—
Longueur du contexte
400K
Modalités
image, text
Radar de capacités
49
general
89
coding
98
reasoning
74
science
70
agents
85
multimodal
Classements
| Domaine | #Rang | Score | Source |
|---|---|---|---|
| Capacité agentique | 68 | 44.0 | LS |
| Classement codage | 84 | 87.0 | AA |
| Classement général | 53 | 74.0 | AA |
| Classement multimodal | 7 | 71.0 | LS |
| Science | 81 | 77.0 | AA |
Scores de benchmarks (LLM Stats)
(LLM Stats (zeroeval))Agents
Toolathlon
46.3%Aut.
Communication
Tau2 Telecom
98.7%Aut.
Tau2 Retail
82.0%Aut.
Language
MMMLU
89.6%Aut.
Math
AIME 2025
100.0%Aut.
HMMT 2025
99.4%Aut.
LiveBench
74.8%
FrontierMath
40.3%Aut.
Multimodal
VideoMMMU
85.9%Aut.
Reasoning
Graphwalks BFS <128k
94.0%Aut.
GPQANYU + Cohere + Anthropic (2023)
92.4%Aut.
BrowseComp Long Context 128k
92.0%Aut.
BrowseComp Long Context 256k
89.8%Aut.
Graphwalks parents <128k
89.0%Aut.
ARC-AGI
86.2%Aut.
CharXiv-R
82.1%Aut.
SWE-Bench Verified
80.0%Aut.
SWE-Lancer (IC-Diamond subset)
74.6%Aut.
BrowseCompOpenAI (2025)
65.8%Aut.
MCP Atlas
60.6%Aut.
ARC-AGI v2
52.9%Aut.
Humanity's Last Exam
34.5%Aut.
Vision
ScreenSpot Pro
86.3%Aut.
MMMU-Pro
79.5%Aut.
Indices d'évaluation AA
(Artificial Analysis)Math Index(Artificial Analysis)99.0
Aime 25(MAA (Mathematical Association of America))99.0
Gpqa(NYU + Cohere + Anthropic (2023))90.3
Livecodebench(UC Berkeley + MIT + Cornell (2024))88.9
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))87.4
Tau2(Sierra + U Toronto + Vector Institute (2025))84.8
Lcr(Artificial Analysis)82.7
Ifbench(Google Research (2023))75.4
Terminalbench Hard(Stanford × Laude Institute (2026))47.0
Hle(Center for AI Safety + Scale AI (2025))37.7
Intelligence Index(Artificial Analysis)30.4
Scores par catégorie LLM Stats
(LLM Stats (zeroeval))Language90
Physics90
Grounding90
Healthcare90
Biology90
Chemistry90
Communication90
Multimodal80
Reasoning80
Search80
Spatial Reasoning80
Frontend Development80
General80
Math70
Code70
Tool Calling70
Vision70
Agents60
Tarification
Prix d'entrée$1.75 / 1M tokens
Prix de sortie$14 / 1M tokens
Prix mixte (3:1)$4.813 / 1M tokens
Prix de lecture cache$0.175 / 1M tokens
Vitesse
Tokens/sec0.0
Délai du premier token0.00s
Temps de réponse0.00s
Classement des Prix par Fournisseur
Classement des Prix par Fournisseur
2 fournisseurs
Moins cher: OpenAIPlus cher: Neon
FournisseurEntréeSortie
1OpenAIMoins cher
$0
$0.00001
2Neon
$1.75
$14
Comparer les prix entre différents fournisseurs API pour ce modèle.