incomplete turn detected: stopReason=stop payloads=0 | Friends of the Crustacean 🦞🤝 | Page 1

fleet loom Apr 21, 2026, 1:12 PM

#

I'm always getting this error with openrouter models: 11:54:37+00:00 warn agent/embedded {"subsystem":"agent/embedded"} incomplete turn detected: runId=78f69b6f-4a28-4c5d-bb0d-f448c643b66a sessionId=4e32e3de-1b9b-4e30-90e9-843bc0def67e stopReason=stop payloads=0 — surfacing error to user

I tested /auto, /free, openai/gpt-oss-120b:free and meta-llama/llama-3.3-70b-instruct:free.

In the openrouter activity page I notice that there's no request sent to openrouter by openclaw.

I tested sending the same request ("content": "[Thread starter - for context]\nThread 1\n\nHello\n\n<<<EXTERNAL_UNTRUSTED_CONTENT id="3fbd612969ac0204">>>\nSource: External\n---\nUNTRUSTED Discord message body\nHello\n<<<END_EXTERNAL_UNTRUSTED_CONTENT id="3fbd612969ac0204">>>") with curl and openrouter replys to it.

patent cosmosBOT Apr 21, 2026, 1:12 PM

#

prisma egretBOT Apr 21, 2026, 1:12 PM

#

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

patent cosmosBOT Apr 21, 2026, 1:12 PM

#

barren tuskBOT Apr 21, 2026, 1:16 PM

#

This looks like a real OpenRouter integration bug in OpenClaw, not a bad prompt.

payloads=0 means OpenClaw ended the turn with stopReason=stop but built no visible assistant output. I checked the code/changelog, and there were several recent OpenRouter fixes for exactly this class of failure:

openrouter:auto / openrouter:free alias mis-resolution
stale OpenRouter base URL normalization (/v1 vs /api/v1)
streamed reasoning_details / response.output_text getting dropped, which also ends in payloads=0

First thing, can you paste these 4 outputs?

openclaw --version
openclaw models status --probe
openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json
openclaw logs --follow

#

What to look for:

model resolves to openrouter/openrouter/auto or openrouter/openrouter/free → alias bug
base URL is anything other than https://openrouter.ai/api/v1 → stale/bad config
OpenRouter gets the request but OpenClaw still says payloads=0 → likely the recent stream parsing bug
OpenRouter gets nothing at all → auth/profile/baseUrl resolution problem

Also, if you're on 2026.4.14 / 2026.4.15-ish, I would upgrade first. A bunch of OpenRouter fixes landed right after that.

Docs:

If you paste those outputs, I can narrow it down pretty quickly. If you're already on latest and still hitting it, this is probably GitHub-issue territory, not user error.

fleet loom Apr 21, 2026, 1:39 PM

#

ubuntu@instance-20260413-1224:~$ openclaw --version
OpenClaw 2026.4.15 (041266a)
ubuntu@instance-20260413-1224:~$ openclaw infer model run --model openrouter/free --prompt "Reply with exactly: smoke-ok" --json
Error: Model override "openrouter/openrouter/free" is not allowed for agent "main".

barren tuskBOT Apr 21, 2026, 1:40 PM

#

Yep, that’s the smoking gun.

On 2026.4.15, openrouter/free is being resolved as openrouter/openrouter/free, which is wrong. That was a real bug, fixed later.

So for that path:

OpenClaw rewrites the model ref incorrectly
the allowlist rejects it
no provider call happens
OpenRouter activity stays empty

That matches what you’re seeing.

Also, 2026.4.15 is right in the OpenRouter payloads=0 regression window, so you may be hitting two bugs at once:

bad alias resolution for /free and /auto
separate OpenRouter response-parsing bug on some concrete models

#

What I’d do

1) Upgrade first

I’d strongly upgrade off 2026.4.15 before debugging further.

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

2) Until then, avoid `/free` and `/auto`

On that version, those aliases are suspect.

3) Check your exact allowlisted model refs

Run:

openclaw models status

#

openclaw config get agents.defaults.models

If agents.defaults.models is set, infer model run --model ... must match one of those exact refs.

Docs:
https://docs.openclaw.ai/concepts/models

Important detail

This error:

Error: Model override "openrouter/openrouter/free" is not allowed

#

is not your OpenRouter key failing. It’s OpenClaw mangling the alias before it even gets to OpenRouter.

If you paste:

openclaw models status
openclaw config get agents.defaults.models

I can tell you the exact concrete model ref to use as a temporary workaround on 2026.4.15.

fleet loom Apr 21, 2026, 1:44 PM

#

ubuntu@instance-20260413-1224:~$ openclaw config get agents.defaults.models

{
"ollama/qwen3.5:2b": {},
"ollama/qwen3.5:4b": {},
"ollama/qwen3.5:9b": {},
"xai/grok-4": {
"alias": "Grok"
},
"google/gemini-3.1-pro-preview": {
"alias": "gemini"
},
"openrouter/auto": {
"alias": "OpenRouter"
}
}
ubuntu@instance-20260413-1224:~$ openclaw models status

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : openrouter/meta-llama/llama-3.3-70b-instruct:free
Fallbacks (2) : groq/llama-3.3-70b-versatile, google/gemini-3.1-flash-preview
Image model : -
Image fallbacks (0): -
Aliases (3) : Grok -> xai/grok-4, gemini -> google/gemini-3.1-pro-preview, OpenRouter -> openrouter/auto
Configured models (6): ollama/qwen3.5:2b, ollama/qwen3.5:4b, ollama/qwen3.5:9b, xai/grok-4, google/gemini-3.1-pro-preview, openrouter/auto

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (0): -

google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyB-...vhfp88sQ
groq effective=env:gsk_rTh2...oH9vGn2L | env=gsk_rTh2...oH9vGn2L | source=env: GROQ_API_KEY | models.json=gsk_rTh2...oH9vGn2L | source=models.json: ~/.openclaw/agents/main/agent/models.json
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
openrouter effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | openrouter:default=sk-or-v1...6d8ff116
xai effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | xai:default=gsk_rTh2...oH9vGn2L

barren tuskBOT Apr 21, 2026, 1:45 PM

#

Okay, this explains it.

Your config is internally inconsistent right now:

Default model: openrouter/meta-llama/llama-3.3-70b-instruct:free
Configured-model allowlist: only openrouter/auto for OpenRouter

So OpenClaw is trying to use a concrete OpenRouter model, but your allowed/configured model catalog only knows about the alias. On 2026.4.15, that combines badly with the OpenRouter alias bugs.

What’s wrong here

agents.defaults.models is acting as an allowlist/catalog. Yours currently has:

"openrouter/auto": { "alias": "OpenRouter" }

#

but not the concrete model you’re actually using:

"openrouter/meta-llama/llama-3.3-70b-instruct:free"

That’s why manual overrides are weird, and it also makes OpenRouter debugging much messier.

I’d fix it like this

Best fix

Upgrade off 2026.4.15 first.

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

Then fix the allowlist

#

Either:

remove agents.defaults.models entirely, if you do not want an allowlist

or

add the exact concrete OpenRouter refs you want to use

Example shape:

{
  agents: {
    defaults: {
      model: {
        primary: "openrouter/meta-llama/llama-3.3-70b-instruct:free"

#

      },
      models: {
        "openrouter/meta-llama/llama-3.3-70b-instruct:free": {},
        "openrouter/openai/gpt-oss-120b:free": {},
        "openrouter/auto": { alias: "OpenRouter" }
      }
    }
  }
}

Then restart the gateway.

Important detail

For OpenRouter, use the full ref when testing:

#

openrouter/meta-llama/llama-3.3-70b-instruct:free
openrouter/openai/gpt-oss-120b:free

Not the bare unprefixed forms on this version.

After that, test again

openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json

If that still gives payloads=0 after upgrade + allowlist cleanup, then you’re into the separate OpenRouter response-parsing bug, and that’s GitHub-issue territory.

Docs for the allowlist behavior:
https://docs.openclaw.ai/concepts/models

If you want, paste your agents.defaults.model block and I can turn it into a clean working snippet.

fleet loom Apr 21, 2026, 1:53 PM

#

ubuntu@instance-20260413-1224:~$ openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json
13:52:36 [agent/embedded] incomplete turn detected: runId=4ef34e21-3b68-4147-ba8e-bf26a2fcec9c sessionId=4ef34e21-3b68-4147-ba8e-bf26a2fcec9c stopReason=stop payloads=0 — surfacing error to user
{
"ok": true,
"capability": "model.run",
"transport": "local",
"provider": "openrouter",
"model": "meta-llama/llama-3.3-70b-instruct:free",
"attempts": [],
"outputs": [
{
"text": "⚠️ Agent couldn't generate a response. Please try again.",
"mediaUrl": null
}
]
}

barren tuskBOT Apr 21, 2026, 1:54 PM

#

Perfect, that isolates it.

Because the fully qualified concrete model still fails:

openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json

this is not just the /free alias bug anymore.

Conclusion

You’re hitting the separate OpenRouter payloads=0 bug in 2026.4.15.

So at this point I would stop changing config. Your repro is already clean enough.

Next step

#

1) Upgrade OpenClaw

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

2) Re-run this exact command

openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json

If it still fails on latest

Then it’s definitely GitHub issue territory. This matches the recent OpenRouter regressions people have already reported:

Your repro is actually really good, I’d paste exactly this into one of those issues if latest still breaks:

#

openclaw --version
openclaw models status
the exact infer model run command
the payloads=0 output

If you want, after you upgrade, paste the new openclaw --version plus the result of that same infer command and I’ll sanity check it.

fleet loom Apr 21, 2026, 2:20 PM

#

I am on the latest version OpenClaw 2026.4.15 (041266a) but the output is the same:

ubuntu@instance-20260413-1224:~$ openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json
14:19:20 [agent/embedded] incomplete turn detected: runId=4ef34e21-3b68-4147-ba8e-bf26a2fcec9c sessionId=4ef34e21-3b68-4147-ba8e-bf26a2fcec9c stopReason=stop payloads=0 — surfacing error to user
{
"ok": true,
"capability": "model.run",
"transport": "local",
"provider": "openrouter",
"model": "meta-llama/llama-3.3-70b-instruct:free",
"attempts": [],
"outputs": [
{
"text": "⚠️ Agent couldn't generate a response. Please try again.",
"mediaUrl": null
}
]
}

barren tuskBOT Apr 21, 2026, 2:21 PM

#

Yep, thanks, that confirms it.

And I should correct myself a bit: 2026.4.15 is still in the affected window. I was too loose when I said “upgrade first”. The OpenRouter fixes I found landed after 2026.4.15, so this is still a real bug, not your config.

Your repro is clean:

openclaw infer model run --model openrouter/meta-llama/llama-3.3-70b-instruct:free --prompt "Reply with exactly: smoke-ok" --json

and it still ends with:

stopReason=stop
payloads=0
no usable output

So at this point, I’d treat it as confirmed OpenClaw/OpenRouter breakage.

#

Best next move

Please file or comment on a GitHub issue with exactly this repro:

Include these points:

version: OpenClaw 2026.4.15 (041266a)
exact infer command
full output showing payloads=0
that curl to OpenRouter works
that OpenRouter activity shows no request from OpenClaw

That last point matters, because it may mean there are actually two OpenRouter bugs in this version range, not just one.

#incomplete turn detected: stopReason=stop payloads=0

What I’d do

1) Upgrade first

2) Until then, avoid /free and /auto

3) Check your exact allowlisted model refs

Important detail

What’s wrong here

I’d fix it like this

Best fix

Then fix the allowlist

Important detail

After that, test again

Conclusion

Next step

1) Upgrade OpenClaw

2) Re-run this exact command

If it still fails on latest

Best next move

2) Until then, avoid `/free` and `/auto`