An imported workflow contains a model identifier that does not exist on the destination system. The following process separates a transient symptom from a repeatable configuration or compatibility issue without relying on a full reinstall.
Recognize the exact failure
Primary symptom. An imported workflow contains a model identifier that does not exist on the destination system. Record the precise point of failure and whether the same input, account or connected device succeeds in another supported mode. That working comparison establishes the scope of the repair.
What can produce this behavior
The first hypothesis is a mismatch in the failing path. List the exact node names and checkpoint identifiers shown as missing. The second is a lower-level state that the interface may not show: check extra_model_paths.yaml, node requirements and model format compatibility. A third possibility is a feature or version limitation, which should be checked against supported documentation before a reset.
For this AI issue, capture model tag, runtime build, backend choice and resource warnings and compare the same controlled input with a known working model or environment. Preserve the last known working configuration so you can tell whether a successful retry came from an actual correction or a temporary recovery.
A decision table for this case
| What you observe | What it suggests | Focused response |
|---|---|---|
| An imported workflow contains a model identifier that does not exist on the destination system | The failure affects the specific workflow described here; the entire system is not necessarily broken. | List the exact node names and checkpoint identifiers shown as missing. |
| The basic comparison exposes a difference between a working and failing configuration. | A local setting, version, account context or device route deserves attention. | Place properly licensed matching models in the expected model directories and refresh the node/model list. |
| The same failure persists after a reversible correction. | A deeper dependency or compatibility issue may remain. | Check extra_model_paths.yaml, node requirements and model format compatibility. |
Before changing anything: Only obtain models from trusted rights-cleared sources; do not execute unknown custom nodes. Save a screenshot or anonymized log line, record the previous setting, and back up relevant user data. Do not reset every related service just because one test fails.
Choose a recovery route. The Beginner path checks list the exact node names and checkpoint identifiers shown as missing. The Advanced path investigates check extra_model_paths.yaml, node requirements and model format compatibility. They address different evidence; use the path that matches your result.
Step-by-Step Troubleshooting
Choose the Beginner route for reversible checks or Advanced for evidence-driven diagnosis.
Capture the reproducible symptom
An imported workflow contains a model identifier that does not exist on the destination system. Before editing settings, note the exact time, affected account or device, and whether it started after an update, move, hardware change or network change. This is the baseline for the ComfyUI Workflow Reports Missing Checkpoint on a Second Computer guide.
Check list the exact node names and checkpoint identifiers shown
List the exact node names and checkpoint identifiers shown as missing. Observe whether the symptom changes. A clean comparison tells you which part of the process is healthy and avoids unnecessary resets.
Adjust place properly licensed matching models in the expected
Place properly licensed matching models in the expected model directories and refresh the node/model list. Change only this variable, record its former value, and try the failing action once. If there is no improvement, restore the prior configuration before exploring the Advanced route.
Verify the original workflow loads without missing-model warnings
The original workflow loads without missing-model warnings and produces a harmless test output. Repeat using the same account, network, file, title or hardware condition that previously failed. Do not count a different test as confirmation.
Correlate the failure evidence
Collect the error text or state relevant to ComfyUI Workflow Reports Missing Checkpoint on a Second Computer. Record model tag, runtime build, backend choice and resource warnings; anonymize account IDs, public IPs and secret-bearing log fragments before sharing.
Inspect extra_model_paths.yaml, node requirements and model format
Check extra_model_paths.yaml, node requirements and model format compatibility. Record the difference between successful and failed cases and keep the first meaningful error rather than the last cascade message.
Compare a controlled second condition
Where practical, compare the same controlled input with a known working model or environment. If the alternate case works, return to the single configuration difference instead of changing multiple services. Use the vendor-supported method for your current version.
Document the regression test
The original workflow loads without missing-model warnings and produces a harmless test output. Leave the previous value documented for rollback, then monitor one normal operating cycle. If symptoms recur, attach the sanitized evidence to a vendor or administrator support request.
Interpret the results
After the first corrective action, interpret the result carefully. If the original task now works, leave the other settings unchanged. If it still fails, check extra_model_paths.yaml, node requirements and model format compatibility. Compare the same controlled input with a known working model or environment rather than adding random fixes on top of one another.
For escalation, retain model tag, runtime build, backend choice and resource warnings, the original output, and a single before/after comparison. Redact passwords, tokens and personal data. A support agent can then distinguish the same controlled input with a known working model or environment from a general outage without repeating risky actions.
How to prove it is fixed
Pass condition: The original workflow loads without missing-model warnings and produces a harmless test output. The validation must exercise the same path that failed originally. A successful settings screen, single ping, application homepage or idle system is not enough if the original failure appears during a different action.
Repeat the test under the original conditions once after a safe relaunch, reconnect or new session. For intermittent failures, record the time and the difference between a clean result and a failed one instead of claiming success from a single attempt.
If the symptom returns
If the problem persists, restore any test-only settings, preserve the evidence and stop before destructive reset, firmware flash or credential recreation. Give official support the exact symptom, the first failing timestamp and what the comparison established. Only obtain models from trusted rights-cleared sources; do not execute unknown custom nodes.
Prevent recurrence: Document the working correction: Place properly licensed matching models in the expected model directories and refresh the node/model list. Retain the accepted result (the original workflow loads without missing-model warnings and produces a harmless test output) as a future regression test after configuration, software or hardware changes.
Solution Checklist
Questions specific to this fault
How do I know the symptom is actually resolved?
The original workflow loads without missing-model warnings and produces a harmless test output. A generic diagnostic test outside the affected workflow is not a substitute.
Should I reset or reinstall everything first?
No. List the exact node names and checkpoint identifiers shown as missing. Then choose the smallest supported intervention: place properly licensed matching models in the expected model directories and refresh the node/model list. Preserve recoverable settings and user data.
What should I include in a support request?
Provide the exact failing action, product and version, time, non-sensitive log evidence, and this deeper observation: Check extra_model_paths.yaml, node requirements and model format compatibility. Do not include credentials or private customer data.
Disqus hosts the discussion. Comments load automatically as you approach this section and may use cookies.