Codex

Codex is OpenAI's AI coding assistant. It talks to models via the OpenAI Responses API. By pointing Codex at a running Tresor instance, you can keep the Codex client experience while hot-switching the underlying model between Anthropic, DeepSeek, MiniMax, a local llama.cpp server, or any other configured downstream — all from the web UI.

All Codex clients — Codex CLI, the ChatGPT desktop app, and the Codex IDE extension for VS Code — share the same configuration file (~/.codex/config.toml). Configure it once, and the model is available in all three clients.

No need to worry about the different API format underlying model is using, Tresor will auto-translate between the OpenAI Responses API (Codex) to other API formats (e.g., claude-opus-5 using the Anthropic API).

This page shows how to wire Codex to Tresor.

✅ Prerequisites

• Installed Tresor following installation guide.
• Codex CLI installed (or ChatGPT desktop app / Codex VS Code extension).
• Launch Codex at least once so that the ~/.codex directory is created.

1. Configure Tresor

There are two ways to prepare Tresor:

1.1 Option 1: via Web UI

Web UI is more intuitive, but require more manual operations.

1. In "Downstreams" tab

   • Click + Add Provider to add the LLM provider you are about to use.

2. In "Aliases" tab

   • Click + New Alias Group
     • Input Model ID: gpt-.*

       You can alternatively add three different alias groups gpt-.*sol, gpt-.*terra, and gpt-.*luna to achieve more precise mapping control.

     • Checked "Regex Pattern"
     • Select the Downstream and Output Model IDs you added previously.

3. In "Settings" tab

   • Click + Add API Key under "Proxy Authentication" section (e.g, sk-proxy-123). Copy it, and we are about to fill it into experimental_bearer_token in the following section.

1.2 Option 2: via config.yaml

Editing yaml is less intuitive, but require less manual operations.

The following example config.yaml uses 3 downstream providers. Replace them based on your actual providers.

~/.config/tresor/config.yaml

bind_addr: 127.0.0.1:11510
proxy_api_keys:
  - sk-proxy-123
... # Omitted showing other settings here
downstreams:
    - id: llama-cpp
      name: llama.cpp
      base_url: http://127.0.0.1:3000 # change to your actual llama.cpp listening port
      api_formats:
        - openai
        - anthropic
      output_model_ids:
        - qwen3.6:27b-mtp-coder
        - qwen3.6:27b-mtp:instruct
    - id: deepseek
      name: DeepSeek
      base_url: https://api.deepseek.com
      api_formats:
        - openai_responses
      output_model_ids:
        - deepseek-v4-flash
        - deepseek-v4-pro
    - id: minimax
      name: MiniMax
      base_url: https://api.minimaxi.com/anthropic
      api_formats:
        - anthropic
      output_model_ids:
        - MiniMax-M2.5
        - MiniMax-M2.7
        - MiniMax-M3
aliases:
    - input_model_id: gpt-.*sol
      is_regex: true
      announced_names:
        - gpt-5.6-sol
      options:
        - id: alias-minimax-m3
          downstream_id: minimax
          output_model_id: MiniMax-M3
        - id: alias-deepseek-pro
          downstream_id: deepseek
          output_model_id: deepseek-v4-pro
    - input_model_id: gpt-.*terra
      is_regex: true
      announced_names:
        - gpt-5.6-terra
      options:
        - id: alias-llama-cpp-qwen-instruct
          downstream_id: llama-cpp
          output_model_id: qwen3.6:27b-mtp:instruct
        - id: alias-deepseek-flash
          downstream_id: deepseek
          output_model_id: deepseek-v4-flash
        - id: alias-minimax-m2-7
          downstream_id: minimax
          output_model_id: MiniMax-M2.7
    - input_model_id: gpt-.*luna
      is_regex: true
      announced_names:
        - gpt-5.6-luna
      options:
        - id: alias-minimax-m2-5
          downstream_id: minimax
          output_model_id: MiniMax-M2.5
    - input_model_id: gpt-.*
      is_regex: true
      announced_names:
        - gpt-5.5
        - gpt-5.4
      options:
        - id: alias-minimax-m2-5-fallback
          downstream_id: minimax
          output_model_id: MiniMax-M2.5

If a not-announced name like gpt-5.3 requests Tresor, the most precisely matched alias will still handle it (in this case, gpt-.*). Optional announced names does not limit the capability of regex alias group's model name matching.

Restart Tresor to apply the changes in yaml.

2. Configure Codex

Manual Setup

(Optional) Login to Your OpenAI Account

If you login before changing config.toml, you will be able to use remote control through the same account after changing config.toml.

Edit config.toml

~/.codex/config.toml

model_provider = "tresor"
model = "gpt-5.6-sol"
model_reasoning_effort = "high"

[model_providers.tresor]
name = "tresor"
base_url = "http://127.0.0.1:11510/v1"
experimental_bearer_token = "sk-proxy-123"
wire_api = "responses"
requires_openai_auth = true

3. Launch Codex

Once configured, all three Codex clients read the same ~/.codex/config.toml — no per-client configuration is needed:

• Codex CLI: enter your project directory and run the codex command.

  cd /path/to/my-project
  codex

• ChatGPT desktop app: shares the same configuration as Codex CLI; simply install the app and start using it.

• Codex IDE extension for VS Code: shares the same configuration as Codex CLI; simply install the extension and start using it.

Try to send a message. If you see the request in Tresor's Logs tab, the configuration is in effect.

Session History After Switching Providers

If your previous sessions seem to be missing after switching to Tresor, don't worry — nothing is deleted. Codex stores session history in separate groups by login method: sessions created with an official ChatGPT subscription and sessions created with a third-party API (such as Tresor) are kept apart, and only the group matching the current configuration is shown. Restoring the previous configuration brings the earlier sessions back, while the Tresor sessions become hidden in turn. Restart the ChatGPT client after switching for the change to take effect.

5. Hot-switch from the web UI

Open http://127.0.0.1:11510 in a browser, go to the Aliases tab, and click any sibling option card to make it active. Codex's next request will be routed to the new downstream — no restart of either Tresor or Codex is required. 🎉

For example, if you have multiple downstreams configured (DeepSeek, a local llama.cpp server, etc.), flip between them with a single click from the web UI, and the agent immediately starts using the new model on the following turn.

🔧 Troubleshooting

Symptom	Likely cause	Fix
401 Unauthorized from Tresor	proxy_api_keys is set but experimental_bearer_token in config.toml doesn't match	Set experimental_bearer_token to one of the keys in proxy_api_keys.
404 model not found from Tresor	"model" in ~/.codex/config.toml doesn't match any alias.	Verify "model" can match at least one alias.

📚 See also

• DeepSeek guide for Codex