Codex CLI default model is now GPT-6.1 Sol: what 0.161.0 changed
Codex CLI rust-v0.161.0, published 2026-10-07, moved the default model in the bundled and Amazon Bedrock catalogs to GPT-6.1 Sol, and the same release brings /mcp login. The QCode setup is in the section below.
Updated 2026-10-19
Point this tool at the QCode endpoint — one key for Claude, GPT and Chinese models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
What the release changed and how to connect QCode
GPT-6.1 Sol since 0.161.0
The release states: "GPT-6.1 Sol is now the default model in the bundled and Amazon Bedrock catalogs." That is the default entry in the bundled catalog and in the Amazon Bedrock catalog, not that this is the only model left.
The next release drops the default label
rust-v0.162.0, published 2026-10-08, carries #51235, which removes the default model label from the TUI model pickers.
Signing in to MCP servers from the terminal
The same release also states that you can sign in to MCP servers from an active terminal session with /mcp login, followed by the name of the server.
The configuration from the official docs
According to the QCode Codex quick start, the top of config.toml carries model_provider = "qcode", model = "gpt-6.1-sol" and model_reasoning_effort = "high"; the [model_providers.qcode] block carries name = "qcode", base_url = "https://api.qcode.cc/openai", wire_api = "responses" and requires_openai_auth = true. Put the key in ~/.codex/auth.json under "OPENAI_API_KEY" as a key that starts with cr_; to use an environment variable instead, replace requires_openai_auth = true with env_key = "QCODE_API_KEY" and export that variable. There are two ways, and you pick one: the official documentation states that Codex ignores env_key when requires_openai_auth = true.
What you see after the default changes
This release only changes who the default is; the model list itself is unchanged, and the details are above. The QCode docs write model = "gpt-6.1-sol", and the command line form is codex --model gpt-6.1-sol.
Official releases and QCode docs as of 2026-10-19
Checked on 2026-10-19: Codex CLI rust-v0.161.0 was published on 2026-10-07 and rust-v0.162.0 on 2026-10-08; what each release changed is above.
The releases and the dates
Codex CLI rust-v0.161.0 is published (2026-10-07): this release only changed the default entry in the catalogs, the details are above.
Codex CLI rust-v0.162.0 is published: the entry #51235 removes the default model label from the TUI model pickers.
The QCode docs write gpt-6.1-sol as the default model, the command line form is codex --model gpt-6.1-sol, and the key goes into ~/.codex/auth.json or the environment variable QCODE_API_KEY.
What the official sources state and what they leave open
Confirmed (official release and QCode docs)
① the default model becomes GPT-6.1 Sol in the bundled catalog and in the Amazon Bedrock catalog; ② you can sign in to MCP servers from an active terminal session; ③ #51235 removes the default model labels from the TUI model pickers; ④ the QCode config.toml lines (top fields, provider block, key placement) are in the QCode setup section above; ⑤ the docs write that gpt-6.1-sol is the recommended default model.
What the official sources leave open
The official release does not say whether the default model will change again later, and it does not preview related changes in the next version; the official release and changelog are the reference for both.
Follow the default or pin the model
Follow the default
Which entry is the default is decided by the official catalog, the change is above; to run the same model every time, see below.
Pin it in the configuration
Pin the model in the model field of config.toml (the docs give model = "gpt-6.1-sol"), or pick one for a single run with codex --model gpt-6.1-sol and codex -m gpt-6-sol `"Refactor the error handling in the authentication module"`. Then the model of each run is fixed and does not depend on changes in the official catalog.
Steps for setup and troubleshooting
① Check the version: the official details come from rust-v0.161.0 (published 2026-10-07). ② Write config.toml: copy the lines from the QCode setup section above, including name = "qcode" in the provider block. ③ Place the key: "OPENAI_API_KEY" in ~/.codex/auth.json with a key that starts with cr_. ④ Check the model: run it once and look at the model list; the QCode docs name gpt-6.1-sol, gpt-6-sol and gpt-5.6-terra, and /models is the reference. ⑤ If you need other MCP services: the same release lets you sign in from a terminal session. ⑥ To switch the model for one run: codex --model gpt-6.1-sol.
Connecting Codex CLI with a QCode key
According to the QCode pages Codex quick start and Codex complete guide, the GPT models that can be set on QCode are gpt-6.1-sol, gpt-6-sol and gpt-5.6-terra, and the list on /models is the reference. QCode does not currently offer gpt-6-luna, so use another GPT model (gpt-6.1-sol / gpt-6-sol / gpt-5.6-terra). Billing is per token, per-model rates are on /models; 5 concurrent requests, the same models as personal plans. The full config.toml setup is in the QCode setup section above. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 to 90 days.
Frequently asked questions
When did the default model change?
Codex CLI rust-v0.161.0, published 2026-10-07.
I only see one model, is something limited?
It is not a restriction, only the default: the entry in the catalog, and the model list itself is unchanged. The models the QCode docs list are gpt-6.1-sol, gpt-6-sol and gpt-5.6-terra, and /models is the reference.
How do I pin a model?
Two ways: pin the model field in config.toml (the docs give model = "gpt-6.1-sol"), or pick one for a single run with codex --model gpt-6.1-sol and codex -m gpt-6-sol `"Refactor the error handling in the authentication module"`.
Where does the key go?
The QCode docs give two ways: put the key in ~/.codex/auth.json under "OPENAI_API_KEY" as a key that starts with cr_, or write env_key = "QCODE_API_KEY" in the configuration so that Codex reads your own environment variable. Use one of the two ways: the official documentation states that Codex ignores env_key when requires_openai_auth = true. The endpoint and the protocol are in the QCode setup section above.
Is GPT-6.1 Sol available on QCode?
The models the QCode docs list are gpt-6.1-sol, gpt-6-sol and gpt-5.6-terra, and the docs also name gpt-6.1-sol as the recommended default model. The list of what is actually available is on /models.
What else came with 0.161.0?
The release also carries the sign-in command: you can sign in to MCP servers from an active terminal session with /mcp login, followed by the name of the server. That and the new default model shipped in the same release.
Sources
Official sources: the Codex CLI releases (api.github.com/repos/openai/codex/releases/tags/rust-v0.161.0 and rust-v0.162.0, fetched and checked 2026-10-19; rust-v0.161.0 published 2026-10-07, rust-v0.162.0 published 2026-10-08), and the QCode pages Codex quick start and Codex complete guide (docs.qcode.cc, fetched and checked 2026-10-19). On the QCode side: the pay-as-you-go page on qcode.cc in five languages (fetched and checked 2026-10-19).
Run Codex CLI on a pack
Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 to 90 days.
Related reading
Pay-as-you-go Packs
One-time quota packs valid for 7–90 days, no subscription; extra packs add to the same key
Codex CLI Complete Guide
OpenAI Codex CLI installation, configuration, and comparison
Codex model_catalog_json configuration guide
Loaded at startup, replaces the bundled catalog, profile wins; the 0.160.0 explicit provider catalog change and /model troubleshooting.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
This page was checked on 2026-10-19 against the official Codex CLI releases and the QCode docs; the default entry and the model list change with the version, so the official releases and /models are the reference.