#Gpt 5.4 xhigh fast mode

1 messages · Page 1 of 1 (latest)

slender heron
#

Can we make the openclaw use the gpt 5.4 xhigh reasoning fast mode while using the codex auth ? But for openclaw

steep windBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> âś… Mark Solution

strange spearBOT
#

Yes — but the trick is that Codex OAuth auth lives under the openai-codex/* provider, not openai/*.

What to do (Codex OAuth + GPT‑5.4 + xhigh)

  1. Login with Codex OAuth (on the gateway host):
openclaw models auth login --provider openai-codex
# or: openclaw onboard --auth-choice openai-codex
  1. Select the Codex model in OpenClaw (either via the picker or explicit):
  • In chat: /model openai-codex/gpt-5.4
  1. Set reasoning (“thinking”) to xhigh
  • One-off for a message: include /think xhigh inline
  • Or set it for the whole session: send a directive-only message:
    • /think xhigh
#

Docs: https://docs.openclaw.ai/tools/thinking and https://docs.openclaw.ai/providers/openai

“Fast mode” part

OpenClaw doesn’t have a separate “fast mode” toggle for Codex OAuth beyond transport + the fact that xhigh will inherently be slower (bigger reasoning budget).

If what you mean is “lower latency streaming”, set transport to WebSocket:

{
  agents: {
    defaults: {
      model: { primary: "openai-codex/gpt-5.4" },
      thinkingDefault: "xhigh",
      models: {
        "openai-codex/gpt-5.4": { params: { transport: "websocket" } }
      }
    }