AIToday
Large Language ModelsAI Business & IndustryASCII.jpPublished: Nov 18, 2026, 19:01 JST

FlashLabs adds live trace export to OrcaRouter

Send on LINE
FlashLabs adds live trace export to OrcaRouter

3 Key Points

  1. What happened

    FlashLabs enhanced the Trace Export log feature in its OrcaRouter gateway so AI request traces stream in real time to Grafana, Datadog, Langfuse, and LangSmith, with no SDK or code changes.

  2. Why it matters

    Teams running AI agents in production can see latency, cost, errors, and routing results in tools they already use, without adding code to send traces.

WHO IT HITSEngineering and operations teams running generative AI applications in production benefit most: they can pipe request-level traces into existing dashboards without touching code, which should lower the barrier to monitoring model latency, cost, and errors.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

The new capability addresses a practical gap in running AI agents and generative AI applications in production. Until now, sending execution traces to an external monitoring tool required embedding an SDK or changing code, which raised the barrier to adoption. With Trace Export, FlashLabs says users can pick a destination in the admin console, enter connection details such as an endpoint and access token, and start streaming request-level data — including model, provider, token count, cost, latency, status, API key name, and user ID — while prompt and completion text is automatically masked by Data Cloak before transmission.

The design also aims to keep response speed unaffected: traces are sent after each request completes, so the gateway does not slow down the response. Connection information such as destination access tokens is stored encrypted, and a sample rate and test trace sending are configurable.

OrcaRouter itself is developed by US research institute Continuum AI and exclusively offered in Japan by FlashLabs. It provides an OpenAI-compatible API that consolidates access to more than 200 LLMs and generative AI models from providers including Anthropic, OpenAI, Google Gemini, DeepSeek, Grok, and Qwen. The company says it is also working to expand operational features that make model calls more visible, so users can make model selection, routing, and cost-management decisions inside familiar monitoring tools.

FAQ
Does using Trace Export require installing an SDK or changing code?
No. Trace Export requires no SDK installation and no code changes; users select a destination and enter connection details in the admin console.
Which monitoring tools can receive traces?
Grafana, Datadog, Langfuse, and LangSmith are supported, and FlashLabs says the list of supported tools will be expanded going forward.
Can I control how much data is sent?
Yes. A sample rate can be set from 1% to 100%, and a test trace can be sent to check the connection.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleJapan Sport Council, Fujitsu to use skeleton-recognition AI from 2026年11月18日