Codex CLI via Tokonomix

//What is Codex CLI?

Codex CLI is OpenAI’s open-source coding agent for the terminal. It reads your repo, proposes and applies edits, and runs commands in a sandbox. A model provider in config.toml defines how it connects — base URL, wire API and auth — so it isn’t locked to OpenAI’s own endpoint.

• An agentic terminal coder with sandboxed command execution.
• Connects to any OpenAI-compatible endpoint via a custom [model_providers.<id>] table.
• Supports a wire_api of chat (Chat Completions) or responses.
• Open source; configured entirely from ~/.codex/config.toml.

//Why route Codex CLI through Tokonomix?

Codex CLI is excellent, but pinned to one vendor’s models by default. A Tokonomix model_provider opens it to 130+ models across families — and a consensus option for the changes a single model shouldn’t decide alone.

• One provider entry, 130+ models — switch with the top-level model key.
• Use tokonomix-consensus for risky refactors, a fast slug for routine edits.
• Per-call spend caps bound an autonomous coding session.
• Region / origin routing keeps a code-streaming agent compliant.

//Connect Tokonomix in config.toml

Add a [model_providers.tokonomix] table with the Tokonomix base_url, an env_key, and wire_api = "chat" (Tokonomix is Chat Completions-compatible). Then activate it with the top-level model and model_provider keys.

model = "anthropic/claude-sonnet-4-6"
# Opt-in (planning/review only, no tool-calling): model = "tokonomix-consensus"
model_provider = "tokonomix"

[model_providers.tokonomix]
name = "Tokonomix"
base_url = "https://tokonomix.ai/api/v1"
env_key = "TOKONOMIX_API_KEY"
wire_api = "chat"

//Verify it works

• Export TOKONOMIX_API_KEY, save ~/.codex/config.toml, and start Codex CLI.
• Run a small task and confirm it uses the Tokonomix provider and model.
• Check the call and token usage in your Tokonomix dashboard.