Phi-4 Mini Instruct
MicrosoftPhiOpen WeightMIT · Uso Comercial
Descripción
Phi 4 Mini Instruct is a lightweight (3.8B parameters) open model built upon synthetic data and filtered web data, focusing on high-quality reasoning. It supports a 128K token context length and is enhanced for instruction adherence and safety via supervised fine-tuning and direct preference optimization.
Fecha de lanzamiento
2024-02-26
Parámetros
3.8B
Longitud del contexto
—
Modalidades
—
Radar de capacidades
18
general
7
coding
18
reasoning
24
science
15
agents
0
multimodal
Rankings
| Dominio | #Posición | Puntuación | Fuente |
|---|---|---|---|
| Ranking de codificación | 592 | 8.0 | AA |
| Ranking general | 611 | 16.0 | AA |
| Ciencia | 594 | 16.0 | AA |
Puntuaciones de benchmarks (LLM Stats)
(LLM Stats (zeroeval))General
MMLU
67.3%Aut.
TruthfulQA
66.4%Aut.
Multilingual MMLU
49.3%Aut.
Arena Hard
32.8%Aut.
Language
BoolQ
81.2%Aut.
MMLU-Pro
52.8%Aut.
Math
MATH-500
94.6%Aut.
GSM8k
88.6%Aut.
MATH
64.0%Aut.
MGSM
63.9%Aut.
AIMEMAA
57.5%Aut.
Reasoning
ARC-C
83.7%Aut.
OpenBookQA
79.2%Aut.
PIQA
77.6%Aut.
Social IQa
72.5%Aut.
BIG-Bench Hard
70.4%Aut.
HellaSwagAI2 (2019)
69.1%Aut.
Winogrande
67.0%Aut.
GPQANYU + Cohere + Anthropic (2023)
52.0%Aut.
Índices de evaluación AA
(Artificial Analysis)Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))69.6
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))46.5
Gpqa(NYU + Cohere + Anthropic (2023))33.1
Ifbench(Google Research (2023))21.1
Lcr(Artificial Analysis)15.3
Livecodebench(UC Berkeley + MIT + Cornell (2024))12.6
Tau2(Sierra + U Toronto + Vector Institute (2025))8.2
Math Index(Artificial Analysis)6.7
Aime 25(MAA (Mathematical Association of America))6.7
Intelligence Index(Artificial Analysis)6.3
Hle(Center for AI Safety + Scale AI (2025))4.4
Coding Index(Artificial Analysis)3.8
Aime(MAA (Mathematical Association of America))3.0
Terminalbench V2 10.4
Terminalbench Hard(Stanford × Laude Institute (2026))0.0
Puntuaciones por categoría LLM Stats
(LLM Stats (zeroeval))Math70
Psychology70
Reasoning70
General70
Language60
Legal60
Finance60
Healthcare60
Physics50
Creativity50
Chat30
Biology30
Chemistry30
Writing30
Precios
Precio de entradaGratis
Precio de salidaGratis
Precio mixto (3:1)Gratis
Velocidad
Tokens/seg47.6
Retraso del primer token0.33s
Tiempo hasta la respuesta0.33s
Ranking de Precios por Proveedor
Ranking de Precios por Proveedor
2 proveedores
Más barato: Azure Cognitive ServicesMás caro: Azure
ProveedorEntradaSalida
1Azure Cognitive ServicesMás barato
$0.075
$0.3
2Azure
$0.075
$0.3
Comparar precios entre diferentes proveedores de API para este modelo.