Goblinz.
In your stack.
Your goblin, beyond the chat.
Hermes, OpenCode and your favorite coding tools are next.
curl http://127.0.0.1:3000/api/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "goblinz-local",
"messages": [
{ "role": "user", "content": "Help me explore a new idea." }
]
}'Your tools.
Goblinz behind them.
Bring Goblinz into the agents and editors you already love. These tools support custom OpenAI-compatible providers — the connection we’re building toward.
Hermes
Give your personal agent a goblin to think with.
Provider setup guideOpenCode
Keep your coding workflow. Bring your own intelligence.
Provider setup guideCline
A little goblin energy for your next build.
Provider setup guideAider
From an idea to a diff, right beside your code.
Provider setup guideFor now, try the local API. Agent connections are not available in this preview. It supports simulated text responses without streaming or tool calling. The guides above describe each tool’s provider settings.
Chat Completions
A familiar message format with the local goblinz-local model. Streaming is not available yet.
Pay as you go
Top up from $10. Your API balance stays available with no expiry.
Keys for every project
Create and revoke separate keys from your API workspace.
Your stash. Your usage.
The same rates for chat and API, per million tokens.
Input includes your message and conversation context. Cached tokens use the reduced rate and are never counted twice. Your token allowance depends on how you use it.
Local preview: estimated tokens, no provider cache. Payments are simulated.
Local preview: the endpoint works, responses are simulated and tokens are estimated. No external AI provider is connected.