Avid @openclaw user, but I think the below is the type of frustration pushing people to Hermes / other.
OpenClaw 2026.6.10 migration gotcha hit today:
If agents are migrated to `openai/gpt-5.5`, make sure provider `openai` actually points at the ChatGPT subscription backend.
Routing is provider-level (`baseUrl`/`auth`/`api`), not just model-string-level. If `openai` still points somewhere else, agents can fail or fall back unexpectedly.
Also check:
- live auth may now be in per-agent SQLite, not old JSON profile files
- active sessions can stay pinned to fallback overrides
- verify with a live agent run + provider field after restart, not config validation alone
Hope this saves someone else a few hours.
@steipete
@Jaytel I don't know if I believe you (well at least I don't beleive the loading times of these images would allow for this to be useable)...if you are in fact achieving some type of high fidelity with fast loading, what are you using to do this?
@openclaw version .22:
gateway is healthy after restart, but openai-codex/gpt-5.5 via ChatGPT/Codex transport is repeatedly timing out at 300s against https://t.co/UO1UNECiza. Telegram then shows generic ‘Something went wrong’ / ‘All models temporarily rate-limited’. Looks like Codex backend transport hangs, not gateway/Telegram failure.
@steipete
any thoughts? #openclaw Will test to see if fixed
@openclaw gateway is healthy after restart, but openai-codex/gpt-5.5 via ChatGPT/Codex transport is repeatedly timing out at 300s against https://t.co/UO1UNECQoI. Telegram then shows generic ‘Something went wrong’ / ‘All models temporarily rate-limited’. Looks like Codex backend transport hangs, not gateway/Telegram failure. @steipete any thoughts? #openclaw
@steipete@openclaw Firstly you're a legend, avid openclaw fan and user from almost day 1 and so epic you respond to our tweets! Second, on version 2026.5.22 and will do
I dont get it. My understanding of Bloom Energy:
~53–65% conversion efficiency
~10–15% higher capex than gas engines/turbines
Only ~$8–11/MWh fuel savings at $6 gas
Still mostly powered by natural gas
Converting to hydrogen likely worsens economics materially
So help me understand the “S-tier” hype.
The company looks less like a revolutionary energy platform and more like a premium emergency-power provider benefiting from a temporary AI/grid bottleneck where hyperscalers need megawatts immediately.
If transmission catches up and solar+BESS keeps falling in cost, what’s the long-term moat here?
I dont get it. My understanding of Bloom Energy:
~53–65% conversion efficiency
~10–15% higher capex than gas engines/turbines
Only ~$8–11/MWh fuel savings at $6 gas
Still mostly powered by natural gas
Converting to hydrogen likely worsens economics materially
So help me understand the “S-tier” hype.
The company looks less like a revolutionary energy platform and more like a premium emergency-power provider benefiting from a temporary AI/grid bottleneck where hyperscalers need megawatts immediately.
If transmission catches up and solar+BESS keeps falling in cost, what’s the long-term moat here?