Cerebras

Cerebras inference provider using the OpenAI-compatible Chat Completions API. Known for fast inference on Cerebras wafer-scale hardware.

Setup

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

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

Set the CEREBRAS_API_KEY environment variable, or pass WithAPIKey() directly.

Models

• llama3.1-8b, llama3.1-70b
• llama-3.3-70b

Tested Models

E2E tested (real API calls, 2026-03-15): llama3.1-8b

Unit tested (mock HTTP server): llama-3.3-70b

Usage

model := cerebras.Chat("llama3.1-8b")

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

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.cerebras.ai/v1 endpoint
WithHeaders(h)	map[string]string	Set additional HTTP headers
WithHTTPClient(c)	*http.Client	Set a custom *http.Client

Notes

• Environment variable CEREBRAS_BASE_URL can override the default endpoint.