OpenClaw v2026.5.12-beta.3: Google/Gemini, recovery notes

v2026.5.12-beta.3 is easiest to understand through failure. The official record describes a provider, model, voice, or context change; the sections below spell out what to save, what to interrupt, and what must remain available afterward.

This candidate is still pre-release. Treat provider, channel, and migration behavior as unconfirmed until your own bounded test passes. Treat the version as one input to the decision. Local credentials, enabled plugins, channels, and host support still need their own evidence.

Failure and recovery path

01 / Gateway/OpenAI HTTP

Source signal. honor max_completion_tokens and max_tokens on inbound /v1/chat/completions requests so client-provided token caps reach the upstream provider via streamParams.maxTokens, with max_completion_tokens taking precedence when both are sent.

Check Gateway/OpenAI HTTP with a known prompt and a disposable account. Save the selected route, reasoning setting, latency, and refusal or fallback result so another operator can reproduce it.

02 / Models/OpenAI CLI auth

Source signal. make openclaw models auth login --provider openai start the ChatGPT/Codex account login by default, while --method api-key remains the explicit OpenAI API-key setup path.

Read Models/OpenAI CLI auth as an authorization checkpoint: prove who may act, what the action can reach, and how the rejected attempt is surfaced to the operator.

03 / Google/Gemini

Source signal. normalize retired Gemini 3 Pro Preview ids inside SDK OAuth auth-result default config patches, so helper-built provider auth flows emit google/gemini-3.1-pro-preview for Gemini 3.1 testing.

Treat Google/Gemini as a boundary change. Exercise one permitted request and one refused request with a test identity, then keep both decisions in the log.

04 / Codex harness

Source signal. keep auth-profile-backed media tools such as image_generate available when OpenAI auth lives in the agent's auth-profile store instead of environment variables.

Treat Codex harness as a boundary change. Exercise one permitted request and one refused request with a test identity, then keep both decisions in the log.

The limits of the recovery promise

WhatsApp/install

allow Baileys' pinned libsignal git subdependency under pnpm 11 so source installs and local checks can complete.

For WhatsApp/install, delivery is only proven when identity survives an interruption. Keep a disposable channel, send one request, and check for a single reply in the intended thread.

Run the smallest useful rehearsal

1. Write down the baseline for Gateway/OpenAI HTTP, including the installed version, host, provider, and workspace.
2. Confirm what can be restored, then take a verified backup before applying the candidate package.
3. Exercise the affected channel with synthetic text and note the thread, sender, reaction, or attachment id that comes back.
4. Restart or reconnect the Gateway, then check for one intended reply rather than a duplicate, missing, or misrouted message.
5. Keep the experiment bounded; an unexpected result belongs in the release record before any wider rollout.

openclaw --version
openclaw gateway status
openclaw security audit

Leave a usable maintenance note

Save the command, output, source link, and environment used for this check. Future operators should be able to tell whether a difference belongs to the release or to local provider, channel, or platform state.

Fix details to carry forward

Codex harness. keep auth-profile-backed media tools such as image_generate available when OpenAI auth lives in the agent's auth-profile store instead of environment variables. This is the edge case to include in a regression check, especially when the release is pre-release.

WhatsApp/install. allow Baileys' pinned libsignal git subdependency under pnpm 11 so source installs and local checks can complete. Treat the fix as a boundary condition and retain the source wording beside your local result.

Keep the full v2026.5.12-beta.3 notes with your test result. Read the official release index for neighboring versions and the model provider reference for configuration limits.