docs(providers/openai): note OpenAI Realtime requires Platform credits, not Codex/ChatGPT subscription#77215
Conversation
openclaw-barnacle
Bot
added
docs
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 725a5371a9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| Fix: top up Platform credits at | ||
| [platform.openai.com/account/billing](https://platform.openai.com/account/billing) | ||
| and configure `OPENAI_API_KEY` (or `talk.providers.openai.apiKey` / | ||
| `voice.realtime.openai.apiKey`) with a key from that organization. For chat |
There was a problem hiding this comment.
Update this path because voice.realtime.openai.apiKey is not a recognized config key, so users who follow the note will set a value that OpenClaw never reads. The realtime voice provider consumes plugins.entries.voice-call.config.realtime.providers.openai.apiKey (or OPENAI_API_KEY), so the current doc can leave Voice Call realtime misconfigured even after applying the suggested fix.
Useful? React with 👍 / 👎.
|
Codex review: needs real behavior proof before merge. Summary Reproducibility: not applicable. for a docs-only PR. Source inspection confirms the documented Talk and Voice Call config paths, the registered realtime provider id, and the current-main documentation gap. Real behavior proof Next step before merge Security Review detailsBest possible solution: Land the narrow docs and changelog clarification after the contributor attaches acceptable rendered-docs or live-output proof and the proof check passes. Do we have a high-confidence way to reproduce the issue? Not applicable for a docs-only PR. Source inspection confirms the documented Talk and Voice Call config paths, the registered realtime provider id, and the current-main documentation gap. Is this the best way to solve the issue? Yes for the content direction: the docs/changelog change is the narrow maintainable fix for the linked confusion. The merge path still needs real behavior proof rather than code repair. Acceptance criteria:
What I checked:
Likely related people:
Remaining risk / open question:
Codex review notes: model gpt-5.5, reasoning high; reviewed against 7d5ca3064a51. |
lonexreb
force-pushed
the
docs/76498-realtime-platform-credits
branch
from
725a537 to
08837fc
Compare
Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
lonexreb
force-pushed
the
docs/76498-realtime-platform-credits
branch
from
08837fc to
f954d6c
Compare
|
Addressed [P2] from codex review ( |
…s, not Codex/ChatGPT subscription (openclaw#76498) Realtime voice (Voice Call realtime.provider: openai and Control UI Talk with talk.provider: openai) goes through the public OpenAI Platform Realtime API and needs an OPENAI_API_KEY whose organization has funded Platform credits. Codex/ChatGPT subscription quota does not cover this route, so an install whose openai-codex/* chat works fine can still hit insufficient_quota / 'You exceeded your current quota' on the first Realtime turn. Adds an inline note next to the Realtime row of the OpenAI capability matrix and a follow-up Note block that explains the symptom, the fix (top up Platform credits + configure OPENAI_API_KEY), and why openai-codex is not registered as a realtime voice provider. This was the central confusion in openclaw#76498. Documenting it inline prevents the same fail-then-discover loop for the next user.
lonexreb
force-pushed
the
docs/76498-realtime-platform-credits
branch
from
f954d6c to
cb74204
Compare
openclaw-barnacle
Bot
added
the
triage: needs-real-behavior-proof
1 participant
Summary
Adds an inline clarification to
docs/providers/openai.mdthat OpenAI Realtime voice (used by Voice Callrealtime.provider: "openai"and Control UI Talk withtalk.provider: "openai") goes through the OpenAI Platform Realtime API and requires anOPENAI_API_KEYwhose organization has funded Platform credits. Codex/ChatGPT subscription quota does not cover this route.Why
From the issue: a user with healthy Codex OAuth (where
openai-codex/*chat worked fine) hitYou exceeded your current quotaon the first Realtime turn, even though the ChatGPT/Codex usage dashboard still showed remaining quota. Trying to force Talk toopenai-codexfailed withRealtime voice provider "openai-codex" is not registered. The fix was to top up OpenAI Platform credits and use thatOPENAI_API_KEYfor Talk — but nothing in the docs surfaced that distinction, so the troubleshooting was guesswork.Changes
docs/providers/openai.mdYes (requires OpenAI Platform API credits, not Codex/ChatGPT subscription).<Note>directly under the matrix that:Voice Call realtime.provider: "openai",Control UI Talk with talk.provider: "openai")insufficient_quota/ "You exceeded your current quota" on the first Realtime turn even when chat works)openai-codexis not a registered realtime voice providerOPENAI_API_KEYfrom that organization)CHANGELOG.md— Unreleased > Changes entry credited to me.Test plan
pnpm exec oxfmt --checkon touched files — clean (oxfmt rewrote whitespace in the matrix row, no content change beyond my edit)<Note>JSX block (matches the existing Note blocks earlier in the file), used root-relative inline references where appropriate, no em-dashes inside headings.Fixes #76498.
Real behavior proof
This is a docs-only change to
docs/providers/openai.md. Proof is the rendered output: a user following the new "Configure OpenAI Realtime credits" steps now sees the actual config keys (talk.providers.openai.apiKeyfor Control UI Talk,plugins.entries.voice-call.config.realtime.providers.openai.apiKeyfor Voice Call) instead of the previous nonexistentvoice.realtime.openai.apiKeypath.pnpm exec oxfmt --check docs/providers/openai.mdclean.