How 1routers works
How model routing works inside Codex
1routers sits between Codex Desktop and supported model providers. You choose the model and routing tier; the routing layer adapts the request, selects an available channel, and preserves the response flow back to Codex.
1routers
See how 1routers connects model choice, protocol adaptation, and route selection inside Codex Desktop.
Choice stays with you
1routers does not hide model selection behind a black box. You choose GPT, Claude, Gemini, Grok, DeepSeek, or another supported model for the task at hand.
Select the model
Match reasoning, coding style, speed, or context needs to the task.
Select the tier
Balance route economics and consistency with Economy, Stable, or Official.
Send from Codex
Keep the desktop workflow, tools, context, and streaming response you already use.
Adaptation happens at the routing layer
Model providers expose different request and response conventions. The routing layer normalizes those differences so a Codex workflow can remain coherent across supported models.
Routing is applied before a response begins. If automatic failover is enabled and a route is unavailable, the request can move to another eligible channel without changing the model you selected.
Availability, latency, and pricing vary by model and route. The active product interface remains the source of truth for current options.
The result returns to the same workspace
The response streams back into Codex Desktop, keeping your interaction in one place instead of forcing a separate client for every model provider.