Heabsy

Heabsy is an OpenAI-compatible inference API for open models. Models in the EEA tier run on dedicated GPUs in EEA data centres with zero data retention; models routed through third parties are labelled as such in the catalog.

Setup

go get github.com/zendev-sh/goai@latest

import "github.com/zendev-sh/goai/provider/heabsy"

Set the HEABSY_API_KEY environment variable, or pass WithAPIKey() directly. Get an API key in the Heabsy console; accounts are opened on request at heabsy.com/contacts.

Models

Model IDs are the names returned by GET /v1/models, for example:

• qwen38 (Qwen3.8 27B: 262,144-token context, up to 32,768 output tokens, image input, tool calling, structured output)

See the Heabsy model catalog for the full list and prices.

Tested Models

Unit tested (mock HTTP server): qwen38

Usage

model := heabsy.Chat("qwen38")

result, err := goai.GenerateText(ctx, model, goai.WithPrompt("Hello"))
if err != nil {
    log.Fatal(err)
}
fmt.Println(result.Text)

Reasoning on/off

qwen38 switches reasoning on or off with chat_template_kwargs.enable_thinking. Pass it with goai.WithProviderOptions; unknown keys are sent in the request body as-is:

result, err := goai.GenerateText(ctx, model,
    goai.WithPrompt("Hello"),
    goai.WithProviderOptions(map[string]any{
        "chat_template_kwargs": map[string]any{"enable_thinking": false},
    }),
)

Options

Option	Type	Description
WithAPIKey(key)	string	Set a static API key
WithTokenSource(ts)	provider.TokenSource	Set a dynamic token source
WithBaseURL(url)	string	Override the default https://api.heabsy.com/v1 endpoint
WithHeaders(h)	map[string]string	Set additional HTTP headers
WithHTTPClient(c)	*http.Client	Set a custom *http.Client

Notes

• Image input works with models that accept it, such as qwen38.
• Heabsy has no embeddings endpoint, so the package provides Chat only.
• Environment variable HEABSY_BASE_URL can override the default endpoint.