Saltar al contenido principal

gpt-oss-120b (High)

OpenAIOpen WeightApache 2.0 · Uso Comercial

Descripción

GPT-OSS-120B is an open-weight, 116.8B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized to run on a single H100 GPU with native MXFP4 quantization. The model supports configurable reasoning depth, full chain-of-thought access, and native tool use, including function calling, browsing, and structured output generation. It achieves near-parity with OpenAI o4-mini on core reasoning benchmarks. Note: While referred to as '120b' for simplicity, it technically has 116.8B parameters.

Fecha de lanzamiento
2025-08-05
Parámetros
116.8B
Longitud del contexto
131K
Modalidades
text

Radar de capacidades

4
general
0
coding
60
reasoning
68
scienceest.
70
agents
0
multimodal

Science se estima a partir de las puntuaciones científicas de LLM Stats o del razonamiento cuando no hay benchmarks científicos dedicados.

Rankings

Dominio#PosiciónPuntuaciónFuente
Capacidad agéntica102
42.0
LS
Ranking general69
3.0
AA
Ciencia69
8.0
AA

Puntuaciones de benchmarks (LLM Stats)

(LLM Stats (zeroeval))

Agents

t2-bench63.9%Aut.

Chat

TAU-bench Retail67.8%Aut.

General

MMLU90.0%Aut.

Healthcare

HealthBench57.6%Aut.
HealthBench Hard30.0%Aut.

Instruction Following

IFBench69.5%Aut.

Language

MMMLU83.8%Aut.
MMLU-Pro80.7%Aut.

Math

AIME 202592.5%Aut.
CodeForces0.82 / 3000Aut.

Reasoning

LiveCodeBench v681.9%Aut.
GPQANYU + Cohere + Anthropic (2023)80.9%Aut.
Humanity's Last Exam14.9%Aut.

Índices de evaluación AA

(Artificial Analysis)
Engineering(Artificial Analysis)
11.2
Economics
9.0
Legal
8.1
Finance And Accounting
7.8
Healthcare And Medical
5.8
Strategy And Ops
5.5
Intelligence Index(Artificial Analysis)
3.8

Puntuaciones por categoría LLM Stats

(LLM Stats (zeroeval))
Language
90
Legal
90
Finance
90
Physics
80
Biology
80
Chemistry
80
Chat
70
Reasoning
70
Communication
70
Tool Calling
70
Math
60
General
60
Healthcare
60
Vision
10

Precios

Precio de entrada$0.15 / 1M tokens
Precio de salida$0.6 / 1M tokens
Precio mixto (3:1)$0.07025 / 1M tokens
Precio de lectura caché$0.075 / 1M tokens

Velocidad

Tokens/seg232.8
Retraso del primer token0.73s
Tiempo hasta la respuesta9.32s

Ranking de Precios por Proveedor

Ranking de Precios por Proveedor

63 proveedores

Más barato: NanoGPTMás caro: CloudFerro Sherlock
ProveedorEntradaSalida
1NanoGPTMás barato
$0.03
$0.15
2CoreWeave
$0.03
$0.17
3Kilo Gateway
$0.03
$0.15
4OrcaRouter
$0.03
$0.17
5Kenari
$0.036
$0.2
6DeepInfra
$0.037
$0.17
7Abacus
$0.037
$0.17
8Deep Infra
$0.037
$0.17
9Phase Router
$0.0385
$0.1768
10Runcrate
$0.03885
$0.1785
11Helicone
$0.04
$0.16
12DevPass (LLM Gateway)
$0.0475
$0.2375
13Novita AI
$0.05
$0.25
14SiliconFlow
$0.05
$0.45
15Merge Gateway
$0.05
$0.25
16Crusoe
$0.05
$0.25
17DInference
$0.0675
$0.27
18Moneo
$0.07
$0.28
19AI Power Grid
$0.075
$0.3
20Powerspring
$0.08842
$0.44208
21Cortecs
$0.089
$0.446
22OVHcloud AI Endpoints
$0.09
$0.47
23FastRouter
$0.09
$0.36
24submodel
$0.1
$0.5
25Baseten
$0.1
$0.5
26Vercel AI Gateway
$0.1
$0.5
27SOKKAN Router
$0.1
$0.5
28OpenReason
$0.1055
$0.422
29OpenAIPRINCIPAL
$0.15
$0.6
30OpenRouter
$0.15
$0.6
31Nebius Token Factory
$0.15
$0.6
32FrogBot
$0.15
$0.6
33Groq
$0.15
$0.6
34Together AI
$0.15
$0.6
35Neon
$0.15
$0.6
36Tinfoil
$0.15
$0.6
37ai&
$0.15
$0.6
38GraphN
$0.15
$0.6
39ZeroGPU
$0.15
$0.6
40Keln
$0.15
$0.6
41Phala
$0.15
$0.6
42Klavo AI
$0.1575
$0.6057
43watsonx.ai
$0.159
$0.636
44Scaleway
$0.16823
$0.6729
45AKI.IO
$0.16823
$0.61683
46SCX.ai
$0.17
$0.55
47Charm Hyper
$0.178
$0.68
48IO.NET
$0.188
$0.7
49Inference APIs
$0.2
$0.8
50SambaNova
$0.22
$0.59
51GreenPT
$0.228
$0.798
52evroc
$0.23
$0.92
53Privatemode AI
$0.2311
$0.7511
54Hugging Face
$0.25
$0.69
55Thinking Machines
$0.33
$0.84
56Cerebras
$0.35
$0.75
57Prime Inference
$0.35
$0.75
58AI/ML API
$0.3703
$1.0051
59stepping stone AG
$0.476
$1.904
60STACKIT
$0.53
$0.76
61Regolo AI
$1.1215
$4.7103
62Opper
$1.1622
$4.88124
63CloudFerro Sherlock
$2.92
$2.92

Comparar precios entre diferentes proveedores de API para este modelo.

Fuentes externas