跳转到主要内容

GPT-5.6 Luna (max)

OpenAIGPTProprietary

描述

GPT-5.6 Luna is the fastest and most cost-efficient tier of OpenAI's GPT-5.6 family, designed for cost-sensitive, high-volume workloads. It brings strong capability at OpenAI's lowest GPT-5.6 price, supports max reasoning effort, and has a 1.05M-token context window.

发布日期
2026-07-09
参数规模
上下文长度
1.1M
支持模态
image, pdf, text

能力雷达图

50
general
68
coding
91
reasoning
67
science
70
agents
85
multimodal

排行榜排名

领域#排名分数来源
智能体能力模型榜50
52.0
LS
代码能力榜20
92.0
AA
通用能力榜34
83.0
AA
多模态榜70
43.0
LS
科学能力38
85.0
AA

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Agents

Connectors99.9%自报
Search and Function-Calling89.7%自报
Capture-the-Flag Challenges (Internal)85.2%自报
Toolathlon53.4%自报
LifeSciBench51.2%自报
Internal Research Debugging Evaluation50.8%自报
Agents' Last Exam50.3%自报
SEC-bench Pro48.9%自报
RSI Index41.9%自报
Big Finance Bench36.0%自报
Management Consulting Tasks (Internal)35.4%自报
PostTrainBench Lite29.6%自报
KernelGen 1P22.4%自报
AutomationBench14.9%自报
ExploitGym12.4%自报
GeneBench-Pro10.8%自报
NanoGPT1.7%自报

Code

BenchCAD (with Python tool)73.9%自报
DeepSWE67.2%自报
DeepSWE 1.167.0%
BenchCAD63.1%自报
ExploitBench33.2%自报

General

Artificial Analysis51.0%
GDP.pdf22.7%自报

Healthcare

HealthBench Consensus95.1%自报
HealthBench55.8%自报
HealthBench Professional55.7%自报
HealthBench Hard32.0%自报

Long Context

MRCR v2 (8-needle, 512K-1M)41.3%自报
MRCR v2 (8-needle)41.3%自报

Math

FrontierMath78.6%自报
FrontierMath Tier 4 (v2)58.5%自报

Multimodal

OSWorld 2.045.6%自报

Reasoning

GPQANYU + Cohere + Anthropic (2023)92.3%自报
Terminal-Bench 2.184.7%自报
BrowseCompOpenAI (2025)83.3%自报
Graphwalks BFS >128k81.3%自报
SWE-Bench ProPrinceton NLP (2024)62.7%自报
Graphwalks BFS 1M51.2%自报
FrontierCode 1.139.8%
MedChemBench (Internal)30.4%自报
ARC-AGI-30.2%自报

Vision

MMMU-Pro (with tools)79.5%自报
MMMU-Pro78.4%自报

AA 评测指数

(Artificial Analysis)
Coding Index(Artificial Analysis)
71.4
Intelligence Index(Artificial Analysis)
52.3
Gpqa(NYU + Cohere + Anthropic (2023))
0.9
Terminalbench V2 1
0.8
Lcr(Artificial Analysis)
0.8
Scicode(UIUC + Argonne National Lab (2024))
0.5
Hle(Center for AI Safety + Scale AI (2025))
0.4
Tau Banking
0.3

LLM Stats 分类评分

(LLM Stats (zeroeval))
Physics
90
Biology
90
Chemistry
90
Search
80
Math
70
Spatial Reasoning
70
Tool Calling
70
Multimodal
60
Vision
60
Long Context
50
Reasoning
50
General
50
Healthcare
50
Agents
50
Code
50
Safety
40
Finance
40
Science
30
Systems
20

定价

输入价格$0.2 / 1M tokens
输出价格$1.2 / 1M tokens
混合价格(3:1)$0.45 / 1M tokens
缓存读取价格$0.02 / 1M tokens
缓存写入价格$0.25 / 1M tokens

速度

Tokens/秒132.4
首Token延迟86.96s
首回答延迟86.96s

供应商价格排行

供应商价格排行

2 个供应商

最便宜: OpenAI最贵: Neon
供应商输入输出
1OpenAI最便宜
$0
$0
2Neon
$1
$6

比较该模型在不同 API 供应商之间的定价。

外部链接