跳轉到主要內容

Atria Dawn Preview

Shanghai AI Laboratory開源權重MIT · 商用許可

描述

Atria Dawn Preview is a preview-stage agentic model from Shanghai AI Laboratory (published under the internlm Hugging Face org), built on a 744B-parameter MoE foundation (the model card states it is trained on the GLM-5.2 base). It targets research and engineering workflows requiring sustained environment understanding, tool use, and multi-step task completion across four capability areas the vendor calls Discovery, Creation, Delivery, and Cybersecurity. The Instruct model accepts text-only input (image/PDF input returns an API error) with a 256K-token context window. Weights are released under the MIT license on Hugging Face and ModelScope in BF16 and FP8 formats. It is served via an international API at api.atria-asi.ai (Chat Completions, Messages, and Responses wire formats) and a China-region endpoint; no per-token pricing has been published on either endpoint as of this writing.

發布日期
2026-09-11
參數規模
744.0B
上下文長度
支援模態

能力雷達圖

100
general
70
coding
100
reasoning
85
science估算
60
agents
0
multimodal

缺少專門科學評測時,Science 由 LLM Stats 科學得分或推理能力估算。

排行榜排名

領域#排名分數來源
智慧體能力模型榜2
75.0
LS

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Agents

AutomationBench53.8%自報
Tau3 Banking41.2%自報

Code

CyberGym86.5%自報
MLE-Bench Lite86.2%自報
SkillsBench66.4%自報

General

GDPval-AA1583.00 / 3000自報

Reasoning

BrowseCompOpenAI (2025)92.5%自報
Terminal-Bench 2.178.3%自報
SWE-Bench ProPrinceton NLP (2024)59.6%自報

Search

DeepSearchQA96.0%自報
WideSearch81.9%自報

Tool Calling

BFCL-V477.0%自報

AA 評測指數

(Artificial Analysis)

暫無 AA 評測資料

LLM Stats 分類評分

(LLM Stats (zeroeval))
Legal
100
Finance
100
Reasoning
100
General
100
Agents
100
Safety
90
Search
90
Code
70
Tool Calling
60

定價

暫無定價資料

速度

暫無速度資料

供應商價格排行

暫無提供商資料

外部連結