⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fres | Friends of the Crustacean 🦞🤝 | Page 1

merry cargo Apr 22, 2026, 9:59 AM

#

openclaw not replying on discord. main GLM 5.1 and fallback minimax-m2.7. showing error "⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fresh session.". even if i "/new". it replied with "⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fresh session.".

dusk sonnetBOT Apr 22, 2026, 9:59 AM

#

icy sealBOT Apr 22, 2026, 9:59 AM

#

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

dusk sonnetBOT Apr 22, 2026, 9:59 AM

#

mellow topazBOT Apr 22, 2026, 10:02 AM

#

I’m checking the docs and known model/runtime failure patterns so I don’t guess here.

#

/new not helping is a useful clue, this is probably not a stale session issue.

That exact message is the generic fallback OpenClaw sends when the model run fails underneath, so I’d check model/auth/runtime first, not Discord transport.

Please paste these:

openclaw --version
openclaw status --all
openclaw models status
openclaw channels status --probe

Then in another terminal:

#

openclaw logs --follow

and send one Discord message that fails, then paste the log lines around it.

Also sanity-test the models directly, outside Discord:

openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json
openclaw infer model run --model minimax/MiniMax-M2.7 --prompt "Reply with: ok" --json

If you use MiniMax OAuth/Coding Plan instead of API key, test:

#

openclaw infer model run --model minimax-portal/MiniMax-M2.7 --prompt "Reply with: ok" --json

Two very common gotchas here:

Wrong model ref/provider
- GLM should be zai/glm-5.1
- MiniMax API key should be minimax/MiniMax-M2.7
- MiniMax OAuth should be minimax-portal/MiniMax-M2.7
Auth mismatch/missing provider auth
- openclaw models status will usually show this fast.

If you’re routing GLM/MiniMax through a local OpenAI-compatible backend instead of the bundled zai / minimax providers, there’s one more gotcha: you may need compat.requiresStringContent: true, and sometimes compat.supportsTools: false.

Docs:

#

Paste the outputs and I’ll narrow it down quickly.

merry cargo Apr 22, 2026, 10:07 AM

#

OpenClaw 2026.4.15 (041266a), Item │ Value │
├─────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Version │ 2026.4.15 │
│ OS │ linux 6.18.7-76061807-generic (x64) │
│ Node │ 22.22.2 │
│ Config │ ~/.openclaw/openclaw.json │
│ Dashboard │ http://127.0.0.1:18789/ │
│ Tailscale │ serve · unknown · pop-os.taild2117f.ts.net · https://pop-os.taild2117f. │
│ │ ts.net │
│ Channel │ stable (default) │
│ Update │ pnpm · npm update 2026.4.21 │
│ Gateway │ local · ws://127.0.0.1:18789 (local loopback) · reachable 31ms · auth │
│ │ token · pop-os (192.168.1.34) app 2026.4.15 linux 6.18.7-76061807- │
│ │ generic │
│ Security │ Run: openclaw security audit --deep │
│ Gateway self │ unknown │
│ Gateway service │ systemd installed · enabled · running (pid 1634, state active)

mellow topazBOT Apr 22, 2026, 10:08 AM

#

I’ve got enough to narrow it down a bit more. I’m checking whether 2026.4.15 had any GLM or MiniMax fixes after your version, so I can tell you if updating is likely part of the fix.

#

That helps a lot.

Your gateway is up and reachable, so this doesn’t look like “Discord is disconnected” or “gateway is dead”.

The big clue is your version: 2026.4.15. There was a fix after that for non-frontier models doing a silent stopReason="error" turn with no output, which can show up exactly as the generic:

“⚠️ Something went wrong while processing your request…”

That makes updating first my strongest recommendation here.

1) Update first

openclaw update

Then verify:

#

openclaw health

Docs: https://docs.openclaw.ai/install/updating

2) If it still fails after updating

Please paste these:

openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json
openclaw infer model run --model minimax/MiniMax-M2.7 --prompt "Reply with: ok" --json

If your MiniMax auth is OAuth/Coding Plan instead of API key, use:

#

openclaw infer model run --model minimax-portal/MiniMax-M2.7 --prompt "Reply with: ok" --json

And in another terminal:

openclaw logs --follow

Then send one Discord message that fails, and paste the log lines around it.

3) Double-check model refs

Make sure your configured refs are exactly:

zai/glm-5.1
minimax/MiniMax-M2.7 for API key
minimax-portal/MiniMax-M2.7 for OAuth

#

If you update and it still breaks, send the models status + one failing log snippet and I can narrow it down pretty fast.

merry cargo Apr 22, 2026, 10:20 AM

#

thz0001@pop-os:~$ openclaw health

🦞 OpenClaw 2026.4.21 (f788c88)
I'm basically a Swiss Army knife, but with more opinions and fewer sharp edges.

│
◇
Discord: ok (@DreamGullyAgent) (1154ms)
Telegram: ok (@Genghistelegram_bot) (1132ms)
Agents: main (default), coo, cmo, cto, cro, marcus, nadia, ren, elena, aisha, julien, suki, tom, carlos, meera, viktor, zara, leo, sam, iris, dante, nathan, priya, jake, hannah, mia, omar, lena, troy, finn, grace, zoe, rafael
Heartbeat interval: 6h (main)
Session store (main): /home/thz0001/.openclaw/agents/main/sessions/sessions.json (6 entries)

agent:main:main (30m ago)
agent:main:discord:direct:1473607912713617471 (46m ago)
agent:main:cron:ff2aadf0-c6af-4fac-a27c-0ff593bc818c (408m ago)
agent:main:cron:ff2aadf0-c6af-4fac-a27c-0ff593bc818c:run:d629517d-e60f-406b-a49b-f2f5670f4d4e (409m ago)
agent:main:cron:8fbcd270-ceae-4229-8e17-b216d91f358f (769m ago)

mellow topazBOT Apr 22, 2026, 10:20 AM

#

Nice, update + health look good. That means the gateway and Discord transport are fine now.

So the next layer is almost certainly model/auth/runtime, not Discord itself.

Please paste these next:

openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json
openclaw infer model run --model minimax/MiniMax-M2.7 --prompt "Reply with: ok" --json

If your MiniMax setup is OAuth/Coding Plan, use this instead of the last one:

openclaw infer model run --model minimax-portal/MiniMax-M2.7 --prompt "Reply with: ok" --json

#

And while testing, run:

openclaw logs --follow

Then send one Discord message that fails, and paste the log lines around it.

At this point the likely buckets are:

wrong model ref/provider
- zai/glm-5.1
- minimax/MiniMax-M2.7 or minimax-portal/MiniMax-M2.7
missing/expired auth for zai or minimax
- models status will show that fast

#

model run fails even outside Discord
- infer model run will confirm it immediately

If you paste models status + one failed infer or log snippet, we should be very close.

merry cargo Apr 23, 2026, 6:03 PM

#

thz0001@pop-os:~$ openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json
openclaw infer model run --model minimax/MiniMax-M2.7 --prompt "Reply with: ok" --json

🦞 OpenClaw 2026.4.21 (f788c88)
Open source means you can see exactly how I judge your config.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : minimax-portal/MiniMax-M2.7
Fallbacks (1) : minimax-portal/MiniMax-M2.7
Image model : google/gemini-3-flash-preview
Image fallbacks (0): -
Aliases (1) : minimax-m2.7 -> minimax-portal/MiniMax-M2.7
Configured models (1): minimax-portal/MiniMax-M2.7

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (1): minimax-portal (1)

google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyCl...9RzwmTtc | models.json=AIzaSyAD..._lX5-AzQ | source=models.json: ~/.openclaw/agents/main/agent/models.json
minimax-portal effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=1, token=0, api_key=0) | minimax-portal:default=OAuth
ollama effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | ollama:default=marker(ollama-local) | models.json=OL...EY | source=models.json: ~/.openclaw/agents/main/agent/models.json
zai effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | zai:default=f1907d56...n6KwizXY

OAuth/token status

minimax-portal
- minimax-portal:default ok expires in 365d
  Error: Model override "zai/glm-5.1" is not allowed for agent "main". how to change default to glm 5.1

mellow topazBOT Apr 23, 2026, 6:05 PM

#

Aha, there it is.

Right now GLM is not enabled for your agent at all.

Your output shows:

Default: minimax-portal/MiniMax-M2.7
Fallbacks: minimax-portal/MiniMax-M2.7
Configured models (1): only minimax-portal/MiniMax-M2.7

So when you try zai/glm-5.1, OpenClaw rejects it because agents.defaults.models is acting as an allowlist.

Safest fix

Add GLM to the allowlist, then make it primary:

#

openclaw config set agents.defaults.models '{"zai/glm-5.1":{}}' --strict-json --merge
openclaw config set agents.defaults.model.primary "zai/glm-5.1"
openclaw config set agents.defaults.model.fallbacks '["minimax-portal/MiniMax-M2.7"]' --strict-json

Then verify:

openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json

What happened

You likely thought GLM was the main model, but the actual config currently says MiniMax portal is both primary and fallback.

Docs:

#

If you want, paste the new openclaw models status after those 3 commands and I’ll sanity-check it.

merry cargo Apr 23, 2026, 6:36 PM

#

thz0001@pop-os:~$ openclaw config set agents.defaults.models '{"zai/glm-5.1":{}}' --strict-json --merge
openclaw config set agents.defaults.model.primary "zai/glm-5.1"
openclaw config set agents.defaults.model.fallbacks '["minimax-portal/MiniMax-M2.7"]' --strict-json
error: unknown option '--merge'

🦞 OpenClaw 2026.4.21 (f788c88) — Automation with claws: minimal fuss, maximal pinch.

Config overwrite: /home/thz0001/.openclaw/openclaw.json (sha256 1e69df66d65d798361b3f290f08d5f8ab76da799b4fa29e552e0c80a06f8fd36 -> a2d5e24c92f384ce770fcac7bf7d6c697af5a058d8747f33e216af13f83e66e9, backup=/home/thz0001/.openclaw/openclaw.json.bak)
Updated agents.defaults.model.primary. Restart the gateway to apply.

🦞 OpenClaw 2026.4.21 (f788c88) — Making 'I'll automate that later' happen now.

Config overwrite: /home/thz0001/.openclaw/openclaw.json (sha256 a2d5e24c92f384ce770fcac7bf7d6c697af5a058d8747f33e216af13f83e66e9 -> 8ff5806954db580b2c80b4e41a34c2c74719f08e90bf5dd26d2fdd11a449d424, backup=/home/thz0001/.openclaw/openclaw.json.bak)
Updated agents.defaults.model.fallbacks. Restart the gateway to apply.

mellow topazBOT Apr 23, 2026, 6:36 PM

#

Yep, my bad, your build doesn’t support --merge there.

You’ve now changed primary/fallback, but the allowlist still hasn’t been updated, so the config is half-switched right now.

Run this exact replacement so both models are allowed:

openclaw config set agents.defaults.models '{"zai/glm-5.1":{},"minimax-portal/MiniMax-M2.7":{"alias":"minimax-m2.7"}}' --strict-json
openclaw gateway restart

Then verify:

openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json

#

If that works, test Discord again.

Why this one matters: agents.defaults.models is the allowlist, and right now it’s probably still only permitting MiniMax. So OpenClaw won’t actually use GLM until that map includes zai/glm-5.1.

If you want, paste the new openclaw models status after that and I’ll double-check it.

merry cargo Apr 23, 2026, 6:37 PM

#

thz0001@pop-os:~$ openclaw models status
openclaw infer model run --model zai/glm-5.1 --prompt "Reply with: ok" --json

🦞 OpenClaw 2026.4.21 (f788c88)
I can grep it, git blame it, and gently roast it—pick your coping mechanism.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : zai/glm-5.1
Fallbacks (1) : minimax-portal/MiniMax-M2.7
Image model : google/gemini-3-flash-preview
Image fallbacks (0): -
Aliases (1) : minimax-m2.7 -> minimax-portal/MiniMax-M2.7
Configured models (1): minimax-portal/MiniMax-M2.7

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (1): minimax-portal (1)

google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyCl...9RzwmTtc | models.json=AIzaSyAD..._lX5-AzQ | source=models.json: ~/.openclaw/agents/main/agent/models.json
minimax-portal effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=1, token=0, api_key=0) | minimax-portal:default=OAuth
ollama effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | ollama:default=marker(ollama-local) | models.json=OL...EY | source=models.json: ~/.openclaw/agents/main/agent/models.json
zai effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | zai:default=f1907d56...n6KwizXY

OAuth/token status

minimax-portal
- minimax-portal:default ok expires in 365d
  {
  "ok": true,
  "capability": "model.run",
  "transport": "local",
  "provider": "zai",
  "model": "glm-5.1",
  "attempts": [],
  "outputs": [
  {
  "text": "ok",
  "mediaUrl": null
  },
  {
  "text": "Done. Primary: GLM 5.1, Fallback: MiniMax M2.7.",
  "mediaUrl": null
  }
  ]
  }

mellow topazBOT Apr 23, 2026, 6:39 PM

#

Nice, this is really good news.

That proves GLM itself is fine:

auth is fine
provider is fine
model run is fine

So we’re no longer debugging ZAI/MiniMax. We’re down to either:

the default model resolution / allowlist path, or
a Discord session/runtime issue

Next, test the exact path Discord uses

Run these:

#

openclaw infer model run --prompt "Reply with: ok" --json
openclaw config get agents.defaults.models --json

The first one tests the default model path with no manual --model.
That’s much closer to what Discord is doing.

Then try Discord again

If it still throws the same error, grab live logs while sending one failing message:

openclaw logs --follow

and paste the lines around the failure.

#

One thing that still looks a bit odd in your output is:

Default: zai/glm-5.1
but Configured models (1): minimax-portal/MiniMax-M2.7

So I want to see the raw agents.defaults.models value next. That should tell us whether the allowlist is actually right or whether status/default are out of sync.

You’re close now. The hard part, “is GLM broken?”, is basically ruled out.

merry cargo Apr 23, 2026, 7:06 PM

#

thz0001@pop-os:~$ openclaw infer model run --prompt "Reply with: ok" --json
openclaw config get agents.defaults.models --json
{
"ok": true,
"capability": "model.run",
"transport": "local",
"provider": "zai",
"model": "glm-5.1",
"attempts": [],
"outputs": [
{
"text": "ok",
"mediaUrl": null
}
]
}
{
"minimax-portal/MiniMax-M2.7": {
"alias": "minimax-m2.7"
}
}
thz0001@pop-os:~$

mellow topazBOT Apr 23, 2026, 7:08 PM

#

Nice, we found the weird bit.

What this means:

openclaw infer model run --prompt ... uses your default model, so GLM works there.
agents.defaults.models is still only:
```
{
  "minimax-portal/MiniMax-M2.7": { "alias": "minimax-m2.7" }
}
```
so your allowlist/session-override catalog is stale.

That can absolutely break Discord if that thread/session is carrying a model override. It also explains why /new didn’t save you.

Do this exact fix

#

openclaw config set agents.defaults.models '{"zai/glm-5.1":{},"minimax-portal/MiniMax-M2.7":{"alias":"minimax-m2.7"}}' --strict-json
openclaw gateway restart

Then verify:

openclaw models status

You want to see Configured models (2) including both:

zai/glm-5.1
minimax-portal/MiniMax-M2.7

Then in the Discord thread

#

Send:

/model zai/glm-5.1

and try a normal message again.

If it still blows up after that, run:

openclaw logs --follow

send one failing Discord message, and paste the log lines around it.

Docs:

#⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fres

1) Update first

2) If it still fails after updating

3) Double-check model refs

Safest fix

What happened

Next, test the exact path Discord uses

Then try Discord again

Do this exact fix

Then in the Discord thread