Hermes Agent on 600+ models, with one plugin
Requesty is a native provider in the Hermes plugin catalog. Install it, add a key, and pick Claude, GPT, Gemini or Kimi in the same model picker as every other provider. Long sessions are cached automatically.
hermes plugins install requesty --enable echo 'REQUESTY_API_KEY=your_requesty_api_key' >> ~/.hermes/.env hermes model # choose Requesty, then a model
Set up Hermes Agent in three steps
The plugin is reviewed by Nous and pinned to a verified version, so there is no YAML to write. The same plugin works in the CLI, the TUI and Hermes Desktop.
- 01Install the plugin
Run hermes plugins install requesty --enable. In Hermes Desktop, open Capabilities, then Plugins, search requesty and add it to the agent.
- 02Add your key
Append REQUESTY_API_KEY to ~/.hermes/.env, or paste it under Settings, Providers, API keys in the desktop app.
- 03Pick a model
Run hermes model, choose Requesty, then a model. The list is fetched live from your account and shows models that support tool calling.
hermes --provider requesty -m claude-sonnet-5-5 # or a routing policy from your organization hermes --provider requesty -m policy/reliable-sonnet
Pick any model from the Hermes picker
The picker shows Requesty managed model IDs, which resolve to the best available provider for that model with fallback built in. Any other ID from the Model Library works with -m.
Reasoning levels set with /reasoning are sent as reasoning_effort on models that support it.
off the input price, at most, for tokens served from a provider cache. Requesty caches this agent by default.
of provider failures on the gateway in April 2026 were 429s. Throttling, not downtime, is what stops an agent mid-task.
eventual success on a routing policy in April 2026, against 85.01% for callers pinned to one provider.
Long system prompts, cached from the second turn
Hermes resends its system prompt and tool definitions, often 15,000 tokens or more, on every turn. Requesty recognises Hermes traffic and caches them automatically, so from the second turn they are billed at a fraction of the input price.
- Nothing to configureCaching is on for Hermes by default; watch cached_tokens in analytics
- One trace per sessionThe Hermes session ID is sent as the trace ID, so a conversation groups into one trace
- Fallback on rate limitsPoint Hermes at a fallback policy and a 429 moves to the next model
Hermes Agent with Requesty, answered
How do I set up Hermes Agent with a model provider?
Install the Requesty plugin with hermes plugins install requesty --enable, add REQUESTY_API_KEY to ~/.hermes/.env, then run hermes model and choose Requesty. You need Hermes 0.21.6 or newer; older versions can add Requesty as a custom provider in ~/.hermes/config.yaml.
Which models can Hermes Agent use through Requesty?
Any model your key is approved for. The hermes model picker lists managed IDs that support tool calling, such as claude-sonnet-5-5, gpt-6.1-sol, gemini-3.8-flash and kimi-k3. Pass any other Model Library ID, or a policy, with -m.
What is the best model for Hermes Agent?
It depends on the task and the budget, so avoid hard coding one. Point Hermes at a routing policy such as policy/reliable-sonnet and change the model behind it in the Requesty dashboard without touching Hermes. Auto Routing goes a step further and picks a model per conversation from routes you write in plain language.
How much does Hermes Agent cost to run?
Hermes is open source, so the running cost is model tokens. Through Requesty you pay the provider price plus 5% from a prepaid balance. Automatic caching of the large system prompt lowers the input bill on long sessions, and a monthly cap on the key bounds it.
How do I fix "rate limit reached" in Hermes Agent?
That error is the upstream provider throttling you with an HTTP 429. Requesty does not cap requests per minute, so the fix is a fallback policy: when the first model is throttled, the request moves to the next model in the chain instead of failing.
How do I change the model in Hermes Agent?
Run hermes model and pick again, type /model in the TUI, or click the model name in the Hermes Desktop composer. For a single session, start Hermes with hermes --provider requesty -m followed by the model ID.
Can I keep Hermes Agent traffic in the EU?
Set REQUESTY_BASE_URL=https://router.eu.requesty.ai/v1 in ~/.hermes/.env to send everything through the EU gateway, or pick an @eu variant such as claude-sonnet-5-5@eu to route only that model through the EU.
Does the Requesty plugin work in Hermes Desktop and the messaging gateway?
Yes. Hermes Desktop installs it from Capabilities, then Plugins, and takes the key under Settings, Providers. Telegram, Discord, Slack and the other platforms use whatever default you set with hermes model.
Guides for the other agents you run
The same key and the same dashboard work across every harness, so spend and logs stay in one place.
Run Hermes Agent on Requesty
600+ models, a spend cap on every key, and every session on the record.