Skip to content
Ascenda

How can I see my AI coding work split by model vendor and route?

Short answer

Most tools show the model name per request, if anything. To see which vendors your work runs on, and whether it goes direct, through a cloud provider such as Bedrock, Vertex or Azure, through a gateway or on device, you need the session history read and grouped. Ascenda Flow does this on your Mac.

Why it matters

Engineers increasingly mix vendors: one model for planning, another in the editor, an open-weights model on a gateway or on device. The same model can also arrive by different routes depending on your employer's setup. Knowing the split helps when you compare how work went, and when you need to know whether you're on the route you're meant to be on.

How Ascenda helps

Flow groups your sessions by model vendor and by route:

  • Vendor: Anthropic, OpenAI, Google and the open-weights families, including DeepSeek, Qwen, Kimi, GLM, MiniMax, Mistral and Llama.
  • Route: direct, a cloud provider (Bedrock, Vertex, Azure), a gateway, a custom endpoint, or on device.

Route is read from Claude Code and Codex; vendor and model also come from Cursor and Windsurf. The route is a category, never a hostname or a country.

Next step

See it on your own weeks

Flow is free for macOS. It imports the history your AI tools already keep, so the first read covers weeks you've already worked.

Get Flow for MacBook a Study

Related questions

Does a different model or effort level change how my AI coding goes?

Flow can assign which model or effort level you use on each working day, in pairs, then lay the failed tool calls side by side, so you compare on your own work.

Where does my Claude Code quota go?

Flow reads the model, effort level and permission mode of each Claude Code response, and keeps your weekly and session limit notices, so you can see what uses your quota.

How much of our team's AI agent work goes through the approved model or vendor?

Declare your approved model vendors and routes; Ascenda reports what share of the cohort's agent time went through them, as an aggregate of ten or more.