Model Guide
Hermes Agent + GLM 5.3 Flash: Frontier Scores on a Flash Budget
GLM 5.3 Flash is the model that sat at the top of public leaderboards for a week under the anonymous name ox-alpha. Z.ai revealed it on 26 August 2026 and open-sourced the weights under MIT the same day. This page covers the Hermes Agent side: which provider slot it uses, the exact model IDs, and the two naming traps that stall most setups.
What you are actually selecting
GLM 5.3 Flash is a mixture-of-experts model with roughly 320 billion total parameters and about 18 billion active per token. It is natively multimodal — text, images and video in, text out — carries a 1M-token context window, and its weights are MIT-licensed.
The name is misleading in a way that matters when you pick a model in Hermes. “Flash” normally signals a distilled, cheaper sibling. Here it signals a different architecture trained from a new base, aimed at keeping frontier capability while cutting serving cost. It is not a downgrade from GLM 5.3, and on public leaderboards it was the stronger of the two.
Trap 1: the provider is called zai
Hermes names providers after the vendor that serves the model, not after the model family. Z.ai is the vendor; GLM is the family. So the provider id is zai while the environment variable is GLM_API_KEY. People grep their config for “glm”, find nothing, and conclude the provider is missing.
# ~/.hermes/.env
GLM_API_KEY=your-key-herehermes model
# choose z.ai / GLM, then GLM 5.3 FlashIf Hermes is not installed yet, start with the installation guide.
Regional endpoints
To move to a different host, override the base URL instead of reaching for a custom provider:
# Standard API key
GLM_BASE_URL=https://api.z.ai/api/paas/v4
# GLM Coding Plan key — a different, dedicated endpoint
GLM_BASE_URL=https://api.z.ai/api/coding/paas/v4This keeps the built-in provider integration and changes only the destination. A custom endpoint is the wrong tool for a host change.
Set the override only if you actually need it. Leaving GLM_BASE_URL unset lets the provider resolve its own default, and hardcoding the general endpoint is precisely what breaks a Coding Plan key — the symptom is an authentication or model-not-found error against a key that works fine elsewhere.
Trap 2: the model ID differs by route
The same model has two different identifiers depending on how you reach it, and pasting the wrong one produces a model-not-found error that looks like an access problem.
| Route | Credential | Model ID |
|---|---|---|
| Built-in Z.ai provider | GLM_API_KEY | Selected through hermes model |
| OpenRouter | OPENROUTER_API_KEY | z-ai/glm-5.3-flash |
Do not guess the spelling from a blog post. Copy the identifier from the current catalog of whichever route you chose — Z.ai and OpenRouter version their IDs independently.
Choosing between the two routes
- Direct Z.ai keeps you on the built-in provider, which means provider-aware behaviour and a single place to change hosts. Best when GLM is your main family.
- OpenRouter gives you one key across many models and instant switching, at the cost of an extra hop and OpenRouter's own availability. Best when you compare models often.
What it costs to run continuously
List pricing is $0.15 per million input tokens. A launch promotion halved that, and promotional pricing is exactly the thing you should not build a monthly budget on. An agent that runs around the clock accumulates input tokens from context far faster than from your typing, so the input rate is the number that decides your bill.
Verify it is actually being used
After selecting the model, send one message and confirm the reply is attributed to the model you chose rather than a fallback. Hermes will keep answering on a previously configured provider if the new key fails to load, and a working agent is not proof that your change took effect.
On OpenClaw Launch you can skip the provider plumbing: deploy a managed Hermes Agent, save your Z.ai or OpenRouter key, and pick the model from the dashboard.
Hermes Agent and GLM 5.3 Flash FAQ
How do I set GLM 5.3 Flash as the Hermes Agent model?
Put GLM_API_KEY in ~/.hermes/.env, then run hermes model and pick the Z.ai provider followed by GLM 5.3 Flash. The provider id is zai — searching your config for “glm” will not find it.
Is GLM 5.3 Flash just a smaller GLM 5.3?
No. Despite the naming, Flash is a separate model with its own newly trained base, not a distilled variant of GLM 5.3. It is the model that ran anonymously as ox-alpha on public leaderboards before Z.ai revealed it on 26 August 2026.
Should I use the Z.ai key or route through OpenRouter?
Use the built-in zai provider when you have a Z.ai account — you keep the provider integration and can move hosts with GLM_BASE_URL. Use OpenRouter when you want one key across many models; the model id there is z-ai/glm-5.3-flash.
Can I point Hermes at a regional Z.ai endpoint?
Yes. Set GLM_BASE_URL rather than switching to a custom provider. For a standard API key the documented value is https://api.z.ai/api/paas/v4. A GLM Coding Plan key is different — Z.AI documents a dedicated endpoint at https://api.z.ai/api/coding/paas/v4 for those — so set the override only when you know which key type you hold.
Is the launch price permanent?
No. The list price is $0.15 per million input tokens, and the lower figure widely quoted at launch was a promotional rate. Budget against list, not against the promotion, because a promotion expires while your agent keeps running.