Mercury 3
InceptionProprietary
描述
Mercury 3 is Inception's flagship diffusion-based LLM (dLLM), released 2026-10-19 and described as the most capable diffusion LLM released to date. Inception says it is more than 40x larger in total parameters than Mercury 1 and runs at about 1,068 output tokens per second on NVIDIA GPUs while matching or beating speed-optimized models such as GPT-6 Luna and Claude Haiku 5.5. It has a 1M-token context window, up to 131,072 output tokens, tunable reasoning effort (instant, low, medium, high), tool calling, and structured outputs. Text in, text out. Available on the Inception API, OpenRouter, Vercel AI Gateway, and Baseten.
发布日期
2026-10-19
参数规模
—
上下文长度
1.0M
支持模态
text
能力雷达图
100
general
50
coding
100
reasoning
77
science估算
30
agents
0
multimodal
缺少专门科学评测时，Science 由 LLM Stats 科学得分或推理能力估算。
排行榜排名
| 领域 | #排名 | 分数 | 来源 |
|---|---|---|---|
| 智能体能力模型榜 | 26 | 59.0 | LS |
基准测试分数 (LLM Stats)
(LLM Stats (zeroeval))Agents
SWE-Atlas
51.1%自报
Tau3 Banking
30.2%自报
Chat
Multi-Challenge
60.4%自报
Code
DeepSWE 1.0
46.3%自报
General
GDPval-AA
1237.20 / 3000自报
Instruction Following
IFBench
76.8%自报
Long Context
MRCR v2
75.5%自报
Reasoning
GPQANYU + Cohere + Anthropic (2023)
89.2%自报
AA-LCR
77.7%自报
Science
OmniScience
46.4%自报
Search
DeepSearchQA
71.3%自报
BrowseComp-Plus
55.2%自报
AA 评测指数
(Artificial Analysis)暂无 AA 评测数据
LLM Stats 分类评分
(LLM Stats (zeroeval))Legal100
Finance100
Agents100
Reasoning100
General100
Physics90
Biology90
Chemistry90
Instruction Following80
Long Context80
Search70
Chat60
Communication60
Knowledge50
Science50
Code50
Tool Calling30
定价
输入价格$0 / 1M tokens
输出价格$0 / 1M tokens
混合价格(3:1)$0 / 1M tokens
缓存读取价格$0.01 / 1M tokens
速度
暂无速度数据
供应商价格排行
供应商价格排行
6 个供应商
最便宜: Inception最贵: Kilo Gateway
供应商输入输出
1Inception主要
$0
$0
2NanoGPT
$0.1
$0.35
3OpenRouter
$0.1
$0.35
4Venice AI
$0.1
$0.35
5Vercel AI Gateway
$0.1
$0.35
6Kilo Gateway
$0.2
$0.7
比较该模型在不同 API 供应商之间的定价。