Open-source · Autonomous · OpenAI-compatible

Run Hermes Agent on any of 20,000+ models

Hermes is a fast-growing open-source autonomous coding agent. Point it at ANT with one command and it inherits cross-provider failover, prompt caching, cost routing, and a routing trace for every request — no lock-in to a single provider.

Install & connect — one commandbash
curl -fsSL https://antbase.ai/hermes.sh | bash

Opens your browser to approve, then auto-writes ~/.hermes/config.yaml — zero copy-paste.

How it works

Connected in three steps — nothing to paste

1
Run the command
The installer starts a secure device-authorization flow and opens your browser.
2
Approve in the browser
Sign in and click approve. A CLI key is minted and handed back to the terminal automatically.
3
Start building
Your ~/.hermes/config.yaml is written for you. Run hermes and you are on ANT.
Manual setup

Prefer to configure it yourself?

Create a key at your dashboard and drop this into ~/.hermes/config.yaml. Hermes is OpenAI-compatible, so ANT's /v1 endpoint is a drop-in.

~/.hermes/config.yamlyaml
# ~/.hermes/config.yaml
# base_url is set explicitly so Hermes calls Antbase directly.
model:
  default: ant:coder
  provider: custom
  base_url: https://antbase.ai/v1
  api_key: ant_your_key_here
  context_length: 200000

Always set base_url. When base_url is present, Hermes calls that endpoint directly and ignores the provider field. Without it, a bare provider: custom can fall through to OpenRouter if OPENROUTER_API_KEY is set in your environment.

What you get

One base-URL swap. Everything ANT does, for free.

Automatic failover
If a provider rate-limits or errors mid-run, ANT retries the same request on another provider — Hermes never sees the failure.
Prompt caching passthrough
Long agent context is cached at the provider where supported, so repeated tool loops cost a fraction of full re-reads.
Cost-based routing
ANT picks the cheapest provider that meets the quality bar for each request — no manual model shopping.
Per-request routing traces
Every Hermes call shows which model and provider served it, latency, tokens, and cost — in your dashboard.
$5 free credits
New workspaces start with $5 of credit. Point Hermes at ANT and start building without a card.
Free models, zero cost
Set the default model to ant:free and Hermes runs entirely on the free-tier model pool.
Pick your default

Set any strategy as your default model

Change one line in your config — model.default — to switch how every Hermes request is routed.

ant:coder
Recommended default
Tuned for coding agents — strong tool-use models, balanced cost.
ant:auto
Smart routing
ANT classifies each request and routes to the best fit automatically.
ant:best
Max quality
Always the strongest available model, cost secondary.
ant:fast
Speed first
Lowest-latency models for snappy interactive edits.
ant:free
Zero cost
Runs entirely on the free-tier pool — great for experimenting.
ant:reasoning
Deep thinking
Routes to reasoning models for planning-heavy tasks.
FAQ

Common questions

Get Hermes on 20,000+ models

One command. Zero copy-paste. Automatic failover on every request.

curl -fsSL https://antbase.ai/hermes.sh | bash