Arize AI 每月有 249,028 次造訪,月增率 +3.2%,定價模式為付費、需聯絡報價。OrcaRouter 每月有 975,300 次造訪,成長率 -5.2%,定價模式為免費增值。
| 面向/指標 | ![]() | |
|---|---|---|
| 官方網域 | arize.com | orcarouter.ai |
| 定價模式 | 付費、需聯絡報價 | 免費增值 |
| 每月造訪次數 | 24.9萬 (249,028) | 97.53萬 (975,300) |
| 月流量變化量 | +7668 (+3.2%) | -5.38萬 (-5.2%) |
| 主要市場與市佔 | 🇺🇸 美國 (51.6%) | 🇺🇸 美國 (21.5%) |
| 每月直接造訪 | 10.78萬 | 24.97萬 |
| 使用者評分 | 4.6 | 4.3 |
| 支援平台 | 網站 | 網站 |

面向生產級 AI 的代理可觀測性、評估與持續改進平台
基於 OpenTelemetry 之上的 OpenInference 規範,蒐集代理與 LLM 呼叫的 span、trace 與工作階段,埋點可跨工具重複使用。
在 span、trace 與工作階段三個層級執行評估,對大量生產資料衡量檢索品質、工具呼叫與最終回覆的優劣。
內建於產品中的代理,可檢查链路、推斷根因,並給出供工程師上線前審閱的修復建議。
當需要讓軌跡資料留在自有基礎設施內時,可自行部署這套領先的開源追蹤與評估工具。
OrcaRouter is an OpenAI-compatible AI gateway that grades every prompt and routes it to the best-value model, then layers on failover, guardrails, an agent firewall, and compliance tooling for production AI.
Each prompt is graded and routed to the model that clears your quality bar, with configurable fallback chains that retry before a response starts.
Pay provider prices with $0 per-token added on pay-as-you-go and subscriptions, or bring your own provider keys and credits.
PII shielding, jailbreak and injection blocking, and per-call approval for tool and MCP actions run before you are billed.
Request logs, per-prompt receipts, routing and spend dashboards, and copy-as-cURL make every call auditable and attributable.

免費 / 開源提供具備容錯移轉的開源 Proxy 與 SDK 以整合 100 多個 LLM,但需要自行架設而非託管服務。

Routes requests across frontier models with automatic failover and analytics, serving as a straightforward alternative API proxy.

擅長實驗追蹤與模型登錄檔工作流程,但較少針對即時 LLM 生產可觀測性進行專門設計。
Gives developers a single OpenAI-compatible endpoint for multiple LLMs with failover, but offers fewer advanced routing rules.

免費 / 開源Includes integrated guardrails and simulation tools, but is less mature for large-scale enterprise production telemetry monitoring.
免費 / 開源Includes integrated guardrails and simulation tools, but is less mature for large-scale enterprise production telemetry monitoring.

免費 / 開源具備強大的測試與提示詞評估工具,不過其監控介面較不針對大規模遙測串流。
高度專注於提示詞管理與版本控制,而 Arize 則優先考慮端到端追蹤與機器學習可觀測性。