To configure a custom provider in Hermes Agent, set the main model’s provider to custom, provide the provider’s model identifier and OpenAI-compatible endpoint, and keep the credential in ~/.hermes/.env. Put non-secret settings in ~/.hermes/config.yaml. For an auxiliary task such as vision, put a direct base_url in that task’s block; Hermes routes the task to that endpoint instead of using the provider selector.
Antihost offers VPS plans with one-click Hermes Agent, NVMe storage and a 14-day free trial.
The same pattern works for hosted gateways and local servers such as Ollama, vLLM, or llama.cpp-compatible APIs. The details that usually cause failures are the YAML location, the endpoint path (often /v1), environment-variable substitution, and which credential Hermes is allowed to use.
As an Amazon Associate I earn from qualifying purchases.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsWhat Hermes means by a custom provider
A custom provider is any model service that exposes the API shape expected by Hermes but is not one of Hermes’s named provider integrations. In the main model configuration, Hermes uses provider: custom as the signal that it should connect to your supplied OpenAI-compatible endpoint. The default value is the model name sent to that service, while base_url identifies the API root.
This is different from selecting a built-in provider. A built-in provider can infer its endpoint and credential variable from the provider name; a custom endpoint requires you to state both the URL and the credential source explicitly. Your service must implement the request and response format supported by the Hermes transport you select, and the model name must exist on that service.
#1 Best Overall
- BLUE LIGHT BLOCKING GLASSES: Our blue light glasses for women boast UV400 protection, blocking harmful blue rays from computer screens. With these bluelight glasses women, you can enjoy gaming, reading, or watching TV with no worries of eye strain.
- EYE RELIEF: Our blue light glasses are engineered to reduce blurry vision or strain caused by prolonged screen time. These womens blue light glasses aid in better sleep by preventing blue light from disrupting your natural sleep cycle, helping you fall asleep faster and enjoy a more restful night's sleep, making them the perfect christmas present.
- COMFORTABLE: Crafted from durable material, our trendy blue light glasses for woman or blue light glasses men ensure long-lasting comfort. The high-transmittance lenses accessories, gamer gifts for men, help keep your eyes relaxed. Perfect stocking stuffers, gifts for gamers men, christmas gifts for men, halloween costumes.
- LIGHTWEIGHT: These clear blue light glasses feature a lightweight frame, meaning they’ll stay comfy for long-term wear. Perfect as blue blockers glasses for women or mens blue light glasses, these non-prescription fake glasses for men can be worn throughout the day. The blue light blocking glasses women feature polycarbonate HD lenses, which are anti-reflective and restore true color.
- WORRY-FREE PURCHASE: If you experience any issues, our support team is ready to assist you quickly. These are the perfect costume glasses for cosplay dressing up.
Think of the configuration as three independent decisions: which Hermes model slot you are changing, where requests should go, and which secret should authorize that destination. Keeping those decisions separate makes it easier to run a hosted primary model alongside a local auxiliary model, or to replace one endpoint without disturbing the rest of the agent.
Before editing configuration
Check the two Hermes files
~/.hermes/config.yamlstores non-secret behavior: model selection, endpoints, model names, auxiliary-task settings, and fallback providers.~/.hermes/.envstores user-managed secrets such as API keys. Keep this file readable only by the account running Hermes and do not commit it to source control.
Hermes reads environment values for substitution in YAML. You can also export variables in the process environment, but using the Hermes .env file keeps the configuration available for normal interactive launches without placing a secret directly in YAML.
Before changing a working setup, note which block currently controls the primary model and which blocks control auxiliary tasks. A correct endpoint under auxiliary.vision will not change the primary agent, and a correct top-level model block will not automatically configure vision or another auxiliary function.
Confirm the provider’s API contract
- Ask the provider for its OpenAI-compatible API root. Many services use a path ending in
/v1; others publish a different root. - Obtain the exact model identifier, including case, punctuation, and any repository or namespace prefix required by that service.
- Determine whether the service requires a real API key, accepts a placeholder key for local access, or uses another authentication mechanism supported by Hermes.
- For a vision task, verify that the selected model is multimodal. A text-only model will not satisfy Hermes’s vision requests even if the endpoint is reachable.
Do not infer the model name from a friendly display name. Hermes sends the configured identifier to the endpoint, so a provider’s catalog name, repository path, or namespace may be significant. Likewise, treat the documented API root as an exact value: a reverse proxy may expose /v1, a versioned path, or another prefix that must remain in the URL.
Configure the main Hermes model
The top-level model block controls the model used by the primary agent. Use this sequence to configure it safely.
- Create the secret variable. Add the provider key to
~/.hermes/.envwith a name that describes the service:
CUSTOM_PROVIDER_API_KEY=your-keyDo not add quotation marks unless they are actually part of the key. If the service does not require authentication, consult its documentation before inventing a key; some local servers accept a placeholder value, while others reject an Authorization header.
Recommended Free Tools
- Edit the main model block. Add the custom provider, model identifier, endpoint, and environment substitution to
~/.hermes/config.yaml:
model:
provider: custom
default: my-model
base_url: https://provider.example/v1
api_key: ${CUSTOM_PROVIDER_API_KEY}- Save the exact endpoint. Include the API prefix required by the service. If its documentation specifies
https://provider.example/v1, do not reduce it to the host name. Conversely, do not append another/v1when the provider already includes that path in a documented base URL. - Start Hermes and make a small request. Use a short prompt first. A successful response confirms that YAML parsing, environment substitution, authentication, endpoint routing, and model lookup all work together.
The key point is placement: provider: custom belongs in the top-level model section for the primary agent. Putting it under an unrelated auxiliary block will not change the main model.
Understand the main-model fields
providerselects Hermes’s routing mode. For an arbitrary OpenAI-compatible service, usecustom.defaultis the model identifier Hermes should request from that endpoint. It is not a human-facing nickname unless the server itself uses that nickname.base_urlis the API root. It must include any required version or reverse-proxy prefix.api_keycan contain a literal value, but an environment substitution keeps the secret out of the YAML file.
These fields work together. A valid key cannot compensate for a wrong model name, and a reachable URL cannot compensate for an endpoint that does not implement the request and response shape Hermes expects.
Rank #2
- WINDEX GAMING WIPES: effortlessly clean all your gaming gear and gaming electronics in one easy step with our new Windex Gaming Wipes
- AMMONIA FREE: our ammonia-free formula is safe to use on all of your favorite gaming equipment like consoles, headsets, controllers, screens, keyboards, and more
- PRE-MOISTENED WIPES: each wipe instantly removes dust, dirt, and stubborn smudges and fingerprints leaving a streak-free shine
- ANTI-STATIC: great for cleaning and wiping down your gaming gear and electronics leaving behind no visible residue
- COMPACT RESEALABLE PACKAGE: store your pack of Windex Gaming Wipes anywhere in your home for whenever a quick and easy wipedown is needed
Use environment substitution without leaking keys
Hermes supports ${VAR_NAME} in config.yaml. You can substitute the endpoint as well as the key:
model:
provider: custom
default: my-model
base_url: ${CUSTOM_PROVIDER_BASE_URL}
api_key: ${CUSTOM_PROVIDER_API_KEY}CUSTOM_PROVIDER_BASE_URL=https://provider.example/v1
CUSTOM_PROVIDER_API_KEY=your-keyIf a referenced variable is missing, Hermes leaves the placeholder text in place and logs a warning. That behavior can look like an authentication or DNS failure because Hermes may try to use a literal value such as ${CUSTOM_PROVIDER_API_KEY}. Check the variable name and the environment loaded by the Hermes process before changing provider settings.
Use different variable names for different services. A custom endpoint does not automatically inherit a key intended for another gateway. This separation prevents a key from being sent to the wrong host and makes rotation predictable.
Environment substitution is useful for non-secret values too. Keeping base_url in a variable lets you switch between a staging gateway and a production gateway without rewriting the model block. Regardless of where the value comes from, verify the resolved URL and key in the same shell, service account, or launcher that starts Hermes; a variable visible in one terminal may not be visible to a background process.
Configure an auxiliary task at a specific endpoint
Auxiliary models are configured under auxiliary.<task>, for example auxiliary.vision. A direct task endpoint is the clearest way to route one task to a separate server:
auxiliary:
vision:
base_url: http://localhost:1234/v1
api_key: local-key
model: qwen2.5-vlWhen base_url is present in an auxiliary task or provider block, Hermes uses that endpoint directly and ignores the block’s provider selector for routing. This lets the main agent use a hosted model while vision, embeddings, or another auxiliary function uses a local or specialist server.
Free tools Windows power users keep installed
One-click scans. No signup required.
Why endpoint precedence matters
A task can appear to name one provider while actually sending requests somewhere else if it also has a direct base_url. The direct URL wins. When debugging, inspect the task block first, then the provider name. This precedence is intentional: it is the most explicit way to route an auxiliary task to a particular endpoint and avoids ambiguity when several provider definitions exist.
Choose the task credential deliberately
Hermes first uses the key configured for the task. If no task key is supplied, it can fall back to OPENAI_API_KEY. It does not reuse OPENROUTER_API_KEY to authenticate an unrelated custom endpoint. Set a task-specific api_key or define the environment variable expected by the endpoint instead of assuming another provider’s key will work.
Rank #3
- Ergonomic & Adjustable Design: Fight "text neck" and improve your posture. Designed for comfort, it offers an adjustable height (7.1in-8.5in) (1.4in adjustable range) and angle (5°-85°) Please note: cell phone stand is only 9 inches tall—the length of a standard pencil
- Without blocking subtitles Perfect for office desk Tiktok Face Time Zoom video—this is the ultimate kitchen desk essential for phones! Perfect viewing experience while reducing neck back strain, A must-have for gamers streamers desk, and anyone who values both style and support!
- Sleek & Stable Design fits Otterbox Case: Say goodbye to bent cables, tangles, and tip-overs! LISEN’s stylish adjustable desk phone holder features a built-in charging hole and smart cable management, keeping your phone secure and your workspace clutter-free cool gadgets for men
- ANTI-SLIP DESIGN free hands: The under 10.00 items easy gadgets back and the bottom of this kitchen essentials for women are fully covered by anti-skid silicone, which LISEN stand up desk decor can provide maximum protection device from any scratches and slides
- Your Perfect Viewing Companion:If your product arrives damaged or breaks within 365 days, reach out to LISEN for a quick and hassle-free solution. Struggling with assembly, like tightening the screw? We’ve got your back! Your experience matters to us. (Product including: phone Holder stand*1, screw*1, hexagon screwdriver*1, specification*1)
Make the model fit the task
The endpoint and model must support the payload Hermes sends. For vision, that means a multimodal model and an API implementation that accepts image content in the format used by the selected Hermes transport. A reachable text model is still an invalid vision configuration.
Configuring an auxiliary model does not make every request use it. The task must actually be invoked, and Hermes must find the block under the exact task name it expects. This distinction is important when a configuration looks correct but normal text prompts continue using the primary model.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Use Hermes’s interactive model picker
Hermes provides the hermes model command for interactive provider and model selection. It is useful when you want Hermes to write the ordinary model keys for you or when you are unsure which existing configuration entry to replace.
- Run
hermes modelin the same user environment that will launch Hermes. - Choose the provider or custom option presented by your installed Hermes version.
- Enter the model identifier and endpoint when prompted.
- For an auxiliary model, choose the relevant task rather than replacing the primary model.
- Open
~/.hermes/config.yamlafterward and verify the generated keys, especiallybase_url,api_key, and the model name.
The auxiliary picker persists its choices under the corresponding auxiliary.<task>.* keys. Interactive setup does not remove the need to protect credentials or confirm that the endpoint path is correct.
Use the picker as a convenience, not as a substitute for review. Check whether it wrote a provider selector, a direct endpoint, or both. If a direct base_url is present, that URL controls routing for the auxiliary task even when a provider name is also displayed.
Connect local and self-hosted servers
Hermes documents local OpenAI-compatible integrations for services such as Ollama, vLLM, and llama.cpp-style servers. Use the server’s local address and the model name loaded by that server. A typical Ollama-style address is http://127.0.0.1:11434. When you provide a bare host and port, Hermes adds /v1 automatically for the documented local-provider patterns.
For a server listening on another local port, specify the complete API root when needed:
model:
provider: custom
default: local-model
base_url: http://127.0.0.1:1234/v1
api_key: local-keySome local servers ignore the key, while others require any non-empty value because their OpenAI-compatible middleware expects an authorization header. Follow the server’s own instructions. A placeholder key should never be treated as a security boundary if the server is exposed beyond the local machine.
Rank #4
- Easy Installation: Our gaming lights have been designed to enhance your entertainment experience. Simply position the led light bar next to your TV or game console to create a game room environment with ease. They come with USB power and a 60 inch cord for direct mounting and mobility
- Audio Syncing & Extensive Mode: The built-in high-sensitivity music sensor allows you to sync the TV led lights to your favorite music. Lights react and move to the beat of music, film and game audio, creating an incredible immersive sound and light experience – and all without the need for an additional hub. You can choose from 64+ patterns, including strobe, illusion, jump, energy, gradient, running, chasing, fluttering mode etc. Each mode can be applied to different scenes
- 64+ Chasing IC+RGB & W Effect: The independent control design of IC+RGB and white all-in-one multifunctional rgbw led bar lights monitor backlight allows you to easily switch the led lights for desk between IC+RGB and white to enjoy a variety of lighting effects, you can choose white lighting mode with high brightness for better illumination when working, etc. Choose IC+RGB lighting mode when you need to create an entertaining atmosphere for the light show, 16 million colours of Color changing rgb light bar let you dance to the beat and create a dynamic party atmosphere
- Dynamic and Dimmable Lights: 1%-full brightness free brightness adjustment gives you the flexibility to adjust dimmable led light bar for room decorations or a variety of occasions
- USB Powered: USB light bars for room can be powered by powernank, usb charger, TV, Monitor, laptop PC mainframe, making behind TV led lighting bar easy to install in any location.TV led lights are powered by the TV usb interface / PC mainframe, can be turned on/off synchronously with TV / PC
Local-server prerequisites
- The server must be running before Hermes sends a request.
- The model name in Hermes must match a model currently available to that server.
- The listening address must be reachable from the process running Hermes;
localhostinside a container may refer to the container rather than the host. - The server must implement the endpoints and response fields expected by Hermes’s selected transport.
- Hardware, model downloads, memory, and concurrency limits are your responsibility; Hermes only supplies the client configuration.
For local deployments, distinguish a server problem from a Hermes problem by checking the server’s own model list and logs first. A model-not-found response usually means the identifier is not loaded, while a connection-refused error usually means the process is stopped, listening on another interface, or inaccessible from Hermes’s network namespace.
Hosted custom endpoint versus local endpoint
The configuration fields are similar, but the operational trade-offs differ.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11| Consideration | Hosted custom service | Local or self-hosted service |
|---|---|---|
| Credential source | Provider-specific key stored in ~/.hermes/.env or the process environment |
Often a local placeholder key; use the server’s required value |
| Endpoint | HTTPS API root supplied by the service, commonly ending in /v1 |
Local host and port, such as http://127.0.0.1:11434 or a complete local /v1 URL |
| Data path | Requests leave the machine for the hosted provider | Inference can remain on your network if the server is not exposed externally |
| Operational work | Account, billing, quotas, and network availability | Model downloads, hardware capacity, server uptime, and upgrades |
| Failure recovery | Use a configured fallback provider if the service is unavailable | Check that the local process is running and the model is loaded before invoking fallbacks |
There is no universal cost or performance advantage in the configuration itself. A hosted service shifts hardware and uptime work to the provider but introduces account, billing, quota, and network dependencies. A local service can keep inference on your network, but you must supply the hardware, download and load the model, and keep the server available.
Configure primary fallbacks
Fallbacks for the primary model belong in config.yaml under fallback_providers. There is no environment-variable equivalent for the primary fallback chain.
fallback_providers:
- provider: openrouter
model: anthropic/claude-sonnet-4The newer list form is the recommended shape for new configurations. Hermes keeps the older single fallback_model form for backward compatibility, but a configuration that needs an ordered provider chain should use fallback_providers. Add a fallback only after confirming that its credential and model are configured; otherwise a primary failure will be followed by another predictable failure.
Fallback configuration is independent of the custom provider’s key. Each fallback must be able to authenticate with its own supported credential, and the model identifier must be valid for that provider. Because the chain is YAML-only, changing environment variables alone cannot add or reorder primary fallbacks.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Troubleshoot a custom provider
Hermes reports an invalid or literal key
- Open
~/.hermes/.envand confirm the variable name exactly matches the${...}expression in YAML. - Restart the Hermes process after changing the environment file.
- Look for the warning about an unresolved substitution. A literal placeholder means Hermes never received the secret.
- Ensure the key belongs to the custom endpoint. A key for another gateway is not automatically valid here.
The request reaches the wrong host
- Check whether you edited the top-level
modelblock or an auxiliary task block. - Inspect the final
base_urlfor a duplicated or missing/v1. - For auxiliary tasks, inspect the task’s own endpoint and model rather than only the global provider name.
The server returns a model-not-found error
- Copy the model identifier from the provider’s model list, including any namespace prefix.
- For local inference, verify that the model has been downloaded or loaded by the server.
- Check that the model supports the task. A vision request requires multimodal capability.
The connection is refused or times out
- Confirm that the hosted service is reachable from the machine running Hermes.
- For local services, confirm the process is listening on the configured port and that a container or virtual machine is not changing the meaning of
localhost. - Verify that HTTPS, HTTP, and any required reverse-proxy path match the provider’s documented endpoint.
An auxiliary model is never used
- Confirm the task is enabled and configured under the exact
auxiliary.<task>name Hermes expects. - Run
hermes modelif you used the interactive picker, then inspect the persisted YAML keys. - Test the task with an input that actually invokes it; configuring a vision model does not make every text request use vision.
The primary model fails without recovery
Add a valid fallback_providers list to config.yaml, provide the fallback’s credential, and retry. If the fallback is missing or malformed, Hermes has no configured alternative to select.
The endpoint responds but Hermes rejects the result
Reachability alone is not enough. Confirm that the server implements the OpenAI-compatible request and response fields expected by Hermes’s selected transport. A service can accept a connection and still fail if it returns a different schema, does not support the requested operation, or cannot accept multimodal content for a vision task.
Best Value
- Compatible with Wide Screens - To ensure compatibility with the dual monitor mount, your each monitor must meet three conditions at the same time: First, computer screens size range: 13 to 32 inches. Second, screen weight range: 4.4 to 19.8 lbs. Third, the back of the monitor screen must have VESA mounting holes with a pitch of 75x75mm or 100x100mm.
- Regarding the compatibility with desks - Your desk must meet three conditions at the same time: First, desk material: Only wooden desks are recommended, plastic or glass desks cannot be used. Second, desk thickness range: 0.59" - 3.54". Third, the bottom of the desk should not have any cross beams or panels, as this will interfere with installation. We recommend carefully checking that your desk and monitors meets all above conditions before purchasing.
- Dual C-Clamp Hold - Worried your dual monitors might wobble or slip? Our upgraded base uses a larger platform plus a dual C-clamp structure to lock the dual monitor arm firmly to your desk. Each arm safely keeps your screens steady while you type, click and game—no shaking, no sliding, just a clean and secure setup you can trust every day. It also provides Grommet Mounting installation choice, both options ensure stable and secure fixation for your 0.59" - 3.54" desk.
- Full-Motion Adjustment For Comfortable View - Pull the screen closer when you’re deep in a spreadsheet, push it back to watch videos, or rotate to portrait for coding — moving everything smoothly with just one hand. The monitor stand offers +85°/-50° tilt, ±90° swivel and 360° rotation. Raise your monitor up to 15.75″ to support a healthy sitting posture. Whether you’re working from home, gaming through the night, or switching between video calls and documents, getting the screens to your natural line of sight helps relieve neck, shoulder and back strain so you can stay focused longer with less fatigue.
- Keep Your Desk Organized: By lifting both screens off the desktop, this dual monitor stand opens up valuable space for your keyboard, notebook, docking station or a simple, clutter-free work area. Built-in cable management guides wires along the arms, keeping cords out of sight and out of the way. Enjoy a tidy, modern workstation that looks as good as it feels to use.
Security and maintenance checklist
- Keep API keys in
~/.hermes/.envor an equivalent protected process environment, not in a shared repository. - Use separate variables for separate providers and rotate a key without changing unrelated model settings.
- Review the destination host before sending sensitive prompts, especially when switching from a local endpoint to a hosted service.
- Restrict a local server’s network exposure; a placeholder key does not protect an openly reachable port.
- After upgrading Hermes, recheck the generated YAML and the transport’s endpoint requirements, particularly if you used the backward-compatible
fallback_modelsetting.
When moving a configuration between machines, copy the non-secret YAML separately from the secret environment file and recreate the required variables on the destination. Verify ownership and permissions on the new .env file before launching Hermes.
Final verification
A custom provider is ready when all of these checks pass:
- The primary model is under the top-level
modelkey and usesprovider: custom. - The configured model identifier is recognized by the endpoint.
- The
base_urlincludes exactly the API prefix required by the service. - Every environment placeholder resolves to a value visible to the Hermes process.
- The credential belongs to that endpoint and is not being borrowed from an unrelated provider variable.
- Each auxiliary task has its own correctly named block, endpoint, key, and task-capable model.
- Any fallback provider is listed in
config.yamland can authenticate independently. - A short primary request and, where applicable, an auxiliary-task request both complete successfully.
Run the primary test first, then test each configured auxiliary task separately. This isolates failures and confirms that endpoint precedence, credentials, model capabilities, and fallback behavior match your intended design.
Frequently Asked Questions
Where does provider: custom go in Hermes Agent?
Put it in the top-level model section for the primary agent. The same setting under an auxiliary block will not change the primary model.
Can I route an auxiliary task directly to a different server?
Yes. Add a base_url inside the relevant auxiliary.<task> block. A task-level base_url takes precedence over the provider selector.
Which key does a custom endpoint use?
Use the task’s configured api_key or the key referenced by your environment substitution. For an auxiliary task without its own key, Hermes can fall back to OPENAI_API_KEY; it does not reuse OPENROUTER_API_KEY for an unrelated custom endpoint.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Why does Hermes warn about an unresolved environment variable?
The referenced variable is missing from the environment available to Hermes. Hermes leaves the ${VAR_NAME} text in place, so check the spelling, ~/.hermes/.env, and the process that launches Hermes.
Do local Ollama, vLLM, and llama.cpp servers work with a custom provider?
Yes, when they expose the OpenAI-compatible API shape Hermes expects. Use the local server address and the exact model name it has loaded; a bare host and port receives /v1 automatically for the documented local-provider patterns.
How do I configure a fallback if the custom provider fails?
Add an ordered fallback_providers list to ~/.hermes/config.yaml. The primary fallback chain has no environment-variable equivalent, and each fallback needs its own valid credential and model.
The Bottom Line
Use provider: custom with the exact model name and OpenAI-compatible base_url in the top-level model block, keep secrets in ~/.hermes/.env, and configure auxiliary endpoints directly under their task blocks. Check URL prefixes, credential isolation, model capability, and YAML fallbacks before diagnosing deeper transport problems.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
