Runs onThe places the vendor's own pages say the product runs: your machine, a container environment you run, or a cloud service the vendor runs. It is not a limit.
Models runWhere the vendor's own pages say the product's models run, not a limit: your machine, a container environment you run, the model company's cloud (servers of the company that makes or serves the model), or a cloud provider's model service. For a cloud, read the data and compliance lines of the company running it, linked by its name.
Several providers in one sessionOne session can use models from different providers, for example one agent on one maker's model and another on a second maker's. Switching the single model for the whole session does not count on its own. A tool that is itself a gateway or library routing to the providers its page names counts.
Several endpoints built inThe tool itself can hold more than one provider or address, so nothing has to sit in front of it to reach different providers.
Local and cloud togetherA local model (on your machine or in a container environment you run) and a cloud model (in its model company's cloud or a cloud provider's model service) can both be used in the same session or setup.
Use a local and a cloud model together: How to set it up(learn.chatgpt.com page, read 2026-10-22) Settings the guide names: model_provider, --local-provider, --profile, config.toml.
Hold several endpoints in the product: How to set it up(learn.chatgpt.com page, read 2026-10-22) Settings the guide names: model_provider, base_url, env_key, config.toml.
Works with
LiteLLM: runs Codex on any model through its Python SDK (github.com)
Compared with similar tools
Same role (harness / agent client) at the same layer (Coding agents and assistants). Up to 4 others, in alphabetical order; nothing is ranked.
Product
Several providersOne session can use models from different providers, for example one agent on one maker's model and another on a second maker's. Switching the single model for the whole session does not count on its own. A tool that is itself a gateway or library routing to the providers its page names counts.
Several endpointsThe tool itself can hold more than one provider or address, so nothing has to sit in front of it to reach different providers.
Model on your machineThe tool can use a model running on your own computer, for example through Ollama, LM Studio, oMLX or llama.cpp.
Local and cloudA local model (on your machine or in a container environment you run) and a cloud model (in its model company's cloud or a cloud provider's model service) can both be used in the same session or setup.
Words in the cells: YesA page we read says so., PartlyA page says part of it, for example separate models per agent without saying the providers can differ., Via a proxyOnly with a proxy, or a local server that imitates a cloud provider's interface, in front of the tool., NoA page we read says the product does not support it., ClaimedA vendor page says so on a marketing or overview page, and we found no guide showing how to set it up., Not foundWe read pages for it and found nothing there. It says nothing about whether the product has the feature., n/aThe question does not apply to the product, for example because it is a model server and not a tool that connects to models., BlankNothing is published for it yet.
Data and compliance
Compliance: Set by whoever runs itThe answer belongs to whoever runs the software, because open-source software has no maker running it for you., for their own deployment.
Data: Set by whoever runs itThe answer belongs to whoever runs the software, because open-source software has no maker running it for you., with telemetry not checked yetWe have not read a page of the right kind for this question yet..
Model and where it runs: Run by its operator.
Details: What the project's own pages state, not a compliance determination.
What the maker's own pages said on the dates shown, not a compliance determination.
Also covers
The vendor's own pages present these as parts or modes of this product, so each is described here and has no page of its own.
Codex cloud
OpenAI's hosted, asynchronous counterpart to the openai/codex CLI: tasks run in parallel cloud environments started from chatgpt.com, GitHub, GitLab, Linear or Slack, continue while the person who started them does something else, and end with a diff to review and a pull request to open.
Our description, from the vendor's documentation, as of 2026-09-23.
Compliance: DPA (company-wideThe mark rests on a sentence that names no product, so check on the page whether it covers this one.); stated as not held: HIPAA BAA.
Data: Training use and default retention page could not be readA page that could answer returned no readable text.; zero-retention option and data residency not stated on the pages readWe read a page that could answer and found no sentence that does. It says nothing about what the maker does..
Model and where it runs: Hosted by the maker; model provider not stated on the pages readWe read a page that could answer and found no sentence that does. It says nothing about what the maker does.; where it can run stated on the maker's pageA sentence on the maker's page answers it for this product..
Details: What the maker's own pages state, not a compliance determination, unverified after 2026-12-23.
What the maker's own pages said on the dates shown, not a compliance determination.
GitHub page: codex(the product's own site, from our list of products; not read for these lines)
The readings behind this page
Our summary of the vendor's pages as of the dates shown on each line (2026-10-19 to 2026-10-24); confirm on the vendor's page.
Cloud models
Codex's advanced-config page says it has a built-in provider for Amazon Bedrock, and that a custom provider is set by a base URL and related fields; its example defines an LLM proxy, a local Ollama and Mistral. (learn.chatgpt.com page, read 2026-10-19)
What you set up
Codex lets the user define extra model providers in config and point the model_provider key at one; the page's example file holds a local Ollama provider beside hosted ones. (learn.chatgpt.com page, read 2026-10-19)
For its local-provider mode, Codex lets the user choose the provider for one run or set a default in config; with neither set, the interactive CLI prompts the user to choose and codex exec ends with an error. (learn.chatgpt.com page, read 2026-10-19)
Models from different providers in one session
Local Codex clients let you define custom agents whose model configurations and instructions differ from task to task. (learn.chatgpt.com page, read 2026-10-21)
Plan, policy and billing
Codex's pricing page, answering what happens at usage limits, says all users can also run extra local chats with an API key, charged at standard API rates. (learn.chatgpt.com page, read 2026-10-23)
Codex's pricing page says that with an API key, image generation is priced at API rates and is not counted against the usage limits included with a ChatGPT plan. (learn.chatgpt.com page, read 2026-10-23)
Codex's Amazon Bedrock page sends questions about AWS credentials, quotas and billing, among other topics, to the customer's own AWS administrator or to AWS Support. (learn.chatgpt.com page, read 2026-10-23)
Codex's models page, in its note on the retirement of gpt-5.4 and gpt-5.4-mini, names gpt-6-luna as the replacement for gpt-5.4-mini and says that on Enterprise and Edu, enabling Luna is something an administrator has to do first. (learn.chatgpt.com page, read 2026-10-19)
A separate model for planning and for doing
Codex's subagents page says a subagent takes the model and reasoning effort of its parent agent unless a subagent model or reasoning effort is configured. (learn.chatgpt.com page, read 2026-10-23)
Codex's subagents page points to gpt-6-luna as the model to use for lighter subagent work when speed and lower cost matter. (learn.chatgpt.com page, read 2026-10-23)
Codex's subagents page advises keeping the main agent focused on the requirements, the decisions and the final outputs. (learn.chatgpt.com page, read 2026-10-21)
Codex's subagents page lists a built-in worker agent, described as execution-focused, for implementation and fixes. (learn.chatgpt.com page, read 2026-10-21)
Codex's subagents page says a custom agent carries over permissions, provider and MCP servers from its parent, and a custom agent file cannot change them. (learn.chatgpt.com page, read 2026-10-23)
About the product
OpenAI's Codex README describes the Codex CLI as OpenAI's coding agent that runs locally on the user's own computer. (github.com page, read 2026-10-24)
OpenAI's Codex documentation says Codex can be given coding tasks, from bug investigations to feature builds, to carry out in the cloud. (learn.chatgpt.com page, read 2026-10-24)
Our reading of vendors' pages, as of the dates shown. Confirm on the vendor's page before relying on any line that cites it. To report an error write to report@crinaro.ai