hermes monitor dashboard examples.html
When routing LLM calls through LLM Gateway, the gateway supports automatic provider-level failover. If the active provider for a given model degrades or goes offline, LLM Gateway transparently switches to an alternative provider serving the same model same model, different provider, no manual intervention required.
For example, DeepSeek V3.2 is available via both CanopyWave and Alibaba Cloud on LLM Gateway. If CanopyWave underperforms, LLM Gateway automatically reroutes to Alibaba Cloud in the background with no model change.
Feature request: Surface this provider and failover information in the WebUI so users can see:
-
Which provider is currently serving each model e.g. "DeepSeek V3.2 via CanopyWave" rather than just "DeepSeek V3.2"
-
When a provider failover has occurred a visible indicator or log entry showing that LLM Gateway switched providers, including the timestamp and which provider it switched from and to
-
Open question: If no working provider exists for the requested model, it is unclear whether LLM Gateway falls back to a completely different model entirely. If it does, the WebUI should flag this prominently so the user knows their agent is no longer running on the intended model.*
-
Failover history per agent — so you can see if a particular agent has experienced provider or model instability over the past 24 hours
This would be surfaced in the agent status table and/or the model health panel. The data source would be the LLM Gateway API, which already tracks which provider served each request.
Reference: LLM Gateway DeepSeek model listings shows multiple providers available per model.
Related: (#692) (provider API cost tracking) the 7-day cost chart on the landing page of the Hermes Monitor dashboard (#721) already shows daily cost per provider. Provider failover visibility would feed directly into this chart, allowing the cost bars to reflect which provider served the model on each day.
hermes monitor dashboard examples.html
When routing LLM calls through LLM Gateway, the gateway supports automatic provider-level failover. If the active provider for a given model degrades or goes offline, LLM Gateway transparently switches to an alternative provider serving the same model same model, different provider, no manual intervention required.
For example, DeepSeek V3.2 is available via both CanopyWave and Alibaba Cloud on LLM Gateway. If CanopyWave underperforms, LLM Gateway automatically reroutes to Alibaba Cloud in the background with no model change.
Feature request: Surface this provider and failover information in the WebUI so users can see:
Which provider is currently serving each model e.g. "DeepSeek V3.2 via CanopyWave" rather than just "DeepSeek V3.2"
When a provider failover has occurred a visible indicator or log entry showing that LLM Gateway switched providers, including the timestamp and which provider it switched from and to
Open question: If no working provider exists for the requested model, it is unclear whether LLM Gateway falls back to a completely different model entirely. If it does, the WebUI should flag this prominently so the user knows their agent is no longer running on the intended model.*
Failover history per agent — so you can see if a particular agent has experienced provider or model instability over the past 24 hours
This would be surfaced in the agent status table and/or the model health panel. The data source would be the LLM Gateway API, which already tracks which provider served each request.
Reference: LLM Gateway DeepSeek model listings shows multiple providers available per model.
Related: (#692) (provider API cost tracking) the 7-day cost chart on the landing page of the Hermes Monitor dashboard (#721) already shows daily cost per provider. Provider failover visibility would feed directly into this chart, allowing the cost bars to reflect which provider served the model on each day.