
NOAN is a fact infrastructure platform for agentic companies: it stores your company's verified, versioned facts in one place and serves them to every AI agent, app, and website over REST and MCP. Instead of agents guessing from stale memory or retrieval, they read the current, attributed version of the truth.
| 面向/指標 | ![]() | |
|---|---|---|
| 官方網域 | deepseek.com | getnoan.com |
| 定價模式 | 免費增值 | 付費 |
| 每月造訪次數 | 3.73億 (372,761,591) | 3663 (3,663) |
| 月流量變化量 | +1536.22萬 (+4.3%) | -684 (-15.7%) |
| 主要市場與市佔 | 🇨🇳 中國 (50.5%) | 🇺🇸 美國 (100%) |
| 每月直接造訪 | 2.92億 | 1450 |
| 使用者評分 | 4.8 | 4.4 |
| 支援平台 | 網站 | 網站 |
由 DeepSeek-V4 與 R1 驅動的高效開源權重與託管推理 AI,token 成本極低。
依託稀疏混合專家架構(V4-Pro 為 1.6T 中的 49B 啟用參數),以極低延遲與記憶體開銷處理最多 100 萬 token。
由 DeepSeek-R1 開創並融入 V3.2 與 V4 思考模式,對複雜數學、邏輯與程式碼除錯展示逐步自驗證過程。
快取命中時,輸入 token 成本降至 Flash 的每百萬 0.003 美元與 V4-Pro 的每百萬 0.022 美元,離峰 UTC 時段費率再低 50%。
在 chat.deepseek.com 及 iOS/Android 上免費提供文件上傳、推理模式與網頁搜尋,核心模型無須付費牆。

NOAN is a fact infrastructure platform for agentic companies: it stores your company's verified, versioned facts in one place and serves them to every AI agent, app, and website over REST and MCP. Instead of agents guessing from stale memory or retrieval, they read the current, attributed version of the truth.
Each fact keeps its date, sources, and the person who verified it, so nothing is silently overwritten and agents can show the receipt behind an answer.
Scope one key per agent, app, or site and limit it to the stacks it needs, so the right keys read the right facts and nothing else.
Turn review on for a project or a single key so any write an agent proposes waits for a human to verify it before it can be read.
See your entire fact base as it stood on any past date, so you can answer what was true when a decision was made.

提供巨大的上下文視窗與深度的 Google 生態系整合,但對於純推理任務感覺較不夠專精。
提供巨大的上下文視窗與深度的 Google 生態系整合,但對於純推理任務感覺較不夠專精。

免費 / 開源提供同等級的開源權重基礎模型與多語言效能,但需要自行架構主機或第三方推論設定。
免費 / 開源Enables running open-source models locally on consumer hardware, but lacks DeepSeek's cloud-scale hosted reasoning power.
免費 / 開源Enables running open-source models locally on consumer hardware, but lacks DeepSeek's cloud-scale hosted reasoning power.