Go SDK

Send one chat.completions request from Go with github.com/sashabaranov/go-openai, then reuse the same client shape in your service code.

Minimum working example

Install the client and export your API key:

go get github.com/sashabaranov/go-openai
export GONKAGATE_API_KEY="gp-your-api-key"

Then run one chat.completions request:

package main

import (
    "context"
    "fmt"
    "log"
    "os"
    "time"

    "github.com/sashabaranov/go-openai"
)

func main() {
    apiKey := os.Getenv("GONKAGATE_API_KEY")
    if apiKey == "" {
        log.Fatal("missing GONKAGATE_API_KEY")
    }

    config := openai.DefaultConfig(apiKey)
    config.BaseURL = "https://api.gonkagate.com/v1"

    client := openai.NewClientWithConfig(config)
    ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
    defer cancel()

    resp, err := client.CreateChatCompletion(
        ctx,
        openai.ChatCompletionRequest{
            Model: "qwen/qwen3-235b-a22b-instruct-2507-fp8",
            Messages: []openai.ChatCompletionMessage{
                {
                    Role:    openai.ChatMessageRoleUser,
                    Content: "Say hello from GonkaGate in one sentence.",
                },
            },
        },
    )

    if err != nil {
        log.Fatal(err)
    }

    fmt.Println(resp.Choices[0].Message.Content)
}

Expected result: if this prints text, your Go client is wired correctly.

Use a current model ID from GET /v1/models. The example value above is only illustrative.

What you need before you run it

• Go 1.18+, because that is the minimum version required by github.com/sashabaranov/go-openai
• A gp-... API key stored in GONKAGATE_API_KEY
• Enough available balance for the request
• A current GonkaGate model ID

Go-specific notes

• Set config.BaseURL = "https://api.gonkagate.com/v1" before you build the client.
• Reuse one configured client in your process or service instead of recreating it on every request.
• Use context.WithTimeout(...) on each request so slow calls fail predictably.
• If you need custom proxy or transport behavior, pass a custom http.Client through config.HTTPClient.

Common errors and limits

• If you forget config.BaseURL = "https://api.gonkagate.com/v1", the client still points at the default OpenAI endpoint.
• 401 invalid_api_key usually means the key is missing, malformed, revoked, or tied to an unavailable account state.
• 404 model_not_found means the model ID is no longer available. Refresh it from GET /v1/models.
• context deadline exceeded means your timeout budget is too small for the request path or current network conditions.
• 429 insufficient_quota means the available prepaid USD balance is too low for the request.

See also

• OpenAI SDK compatibility for the shared BaseURL, key format, model-selection rules, and additive usage fields across runtimes.
• Authentication and API Keys for key creation, secure storage, and rotation.
• GonkaGate API Reference Overview when the first Go request works and you need exact request fields, streaming behavior, and failure policy.
• GET /v1/models for the current machine-readable model IDs.
• OpenAI to GonkaGate Migration Guide if you are switching already-working OpenAI code instead of starting from zero.