Перейти к основному содержанию
AITier

GPT-4.1 mini

OpenAIGPTProprietary

Описание

GPT-4.1 mini provides a balance between intelligence, speed, and cost. It's a significant leap in small model performance, even beating GPT-4o in many benchmarks while reducing latency and cost.

Дата выхода
2025-04-14
Параметры
—
Длина контекста
1.0M
Модальности
image, pdf, text

Радар способностей

30
general
31
coding
54
reasoning
48
science
50
agents
85
multimodal

Рейтинги

Домен#МестоОценкаИсточник
Рейтинг кодинга405
33.0
AA
Общий рейтинг345
40.0
AA
Мультимодальный рейтинг77
52.0
LS
Наука410
36.0
AA

Оценки бенчмарков (LLM Stats)

(LLM Stats (zeroeval))

Chat

IFEvalGoogle Research (2023)84.1%Сам.
Multi-IF67.0%Сам.
TAU-bench Retail55.8%Сам.
Multi-Challenge35.8%Сам.

General

MMLU87.5%Сам.
Internal API instruction following (hard)45.1%Сам.
Aider-Polyglot34.7%Сам.
Aider-Polyglot Edit31.6%Сам.

Language

MMMLU78.5%Сам.
COLLIE54.6%Сам.

Long Context

ComplexFuncBench49.3%Сам.
OpenAI-MRCR: 2 needle 128k47.2%Сам.
OpenAI-MRCR: 2 needle 1M33.3%Сам.

Math

MathVista73.1%Сам.
AIME 202449.6%Сам.
AIME 202540.2%Сам.
HMMT 202535.0%Сам.

Multimodal

MMMU72.7%Сам.

Reasoning

CharXiv-D88.4%Сам.
GPQANYU + Cohere + Anthropic (2023)65.0%Сам.
Graphwalks BFS <128k61.7%Сам.
Graphwalks parents <128k60.5%Сам.
CharXiv-R56.8%Сам.
TAU-bench Airline36.0%Сам.
SWE-Bench Verified23.6%Сам.
Graphwalks BFS >128k15.0%Сам.
Graphwalks parents >128k11.0%Сам.
Humanity's Last Exam3.7%Сам.

Индексы оценки AA

(Artificial Analysis)
Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))
92.5
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
78.1
Gpqa(NYU + Cohere + Anthropic (2023))
66.4
Tau2(Sierra + U Toronto + Vector Institute (2025))
52.9
Livecodebench(UC Berkeley + MIT + Cornell (2024))
48.3
Aime 25(MAA (Mathematical Association of America))
46.3
Math Index(Artificial Analysis)
46.3
Lcr(Artificial Analysis)
44.0
Aime(MAA (Mathematical Association of America))
43.0
Ifbench(Google Research (2023))
38.3
Coding Index(Artificial Analysis)
20.2
Intelligence Index(Artificial Analysis)
10.2
Terminalbench V2 1
10.1
Terminalbench Hard(Stanford × Laude Institute (2026))
7.6
Tau Banking
5.4
Hle(Center for AI Safety + Scale AI (2025))
5.0

Оценки категорий LLM Stats

(LLM Stats (zeroeval))
Legal
90
Finance
90
Instruction Following
80
Healthcare
80
Language
70
Multimodal
70
Physics
70
Structured Output
70
Biology
70
Chemistry
70
Chat
60
Vision
60
Math
50
Reasoning
50
General
50
Communication
50
Tool Calling
50
Writing
50
Spatial Reasoning
40
Long Context
30
Code
30
Frontend Development
20

Цены

Цена ввода$0.4 / 1M токенов
Цена вывода$1.6 / 1M токенов
Смешанная цена (3:1)$0.7 / 1M токенов
Цена чтения кэша$0.1 / 1M токенов

Скорость

Токенов/сек0.0
Задержка первого токена0.00s
Время до первого ответа0.00s

Рейтинг цен провайдеров

Рейтинг цен провайдеров

24 провайдеров

Самый дешевый: OpenAIСамый дорогой: Cortecs
ПровайдерВводВывод
1OpenAIСамый дешевый
$0
$0
2Ofox
$0.32
$1.28
3Poe
$0.36
$1.4
4Helicone
$0.4
$1.6
5302.AI
$0.4
$1.6
6NanoGPT
$0.4
$1.6
7Abacus
$0.4
$1.6
8OpenRouter
$0.4
$1.6
9Kilo Gateway
$0.4
$1.6
10SAP AI Core
$0.4
$1.6
11Cloudflare AI Gateway
$0.4
$1.6
12Azure Cognitive Services
$0.4
$1.6
13Vercel AI Gateway
$0.4
$1.6
14DevPass (LLM Gateway)
$0.4
$1.6
15Azure
$0.4
$1.6
16NEAR AI Cloud
$0.4
$1.6
17OrcaRouter
$0.4
$1.6
18Merge Gateway
$0.4
$1.6
19Pioneer
$0.4
$1.6
20Impossibl
$0.4
$1.6
21Eden AI
$0.4
$1.6
22LLM Gateway
$0.4
$1.6
23Aixy
$0.4
$1.6
24Cortecs
$0.434
$1.704

Сравнение цен разных API-провайдеров для этой модели.

Внешние ссылки

LLM StatsArtificial Analysis