Runcrate

Runcrate inference provider using the OpenAI-compatible Chat Completions API. Runcrate serves open models such as DeepSeek, GLM, Qwen, gpt-oss and Kimi.

Setup

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

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

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

Models

• deepseek-ai/DeepSeek-V4.1-Flash
• deepseek-ai/DeepSeek-V4-Pro
• zai-org/GLM-5.3
• zai-org/GLM-5.3-Flash
• Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo
• openai/gpt-oss-120b
• moonshotai/Kimi-K2.6

Model IDs use the org/model prefix format and are case-sensitive. GET https://api.runcrate.ai/v1/models lists the full catalog and needs no API key.

Tested Models

Unit tested (mock HTTP server, 2026-10-27): deepseek-ai/DeepSeek-V4.1-Flash, zai-org/GLM-5.3

Live tested (real API, 2026-10-24): deepseek-ai/DeepSeek-V4.1-Flash (GenerateText), zai-org/GLM-5.3-Flash (StreamText)

Usage

model := runcrate.Chat("deepseek-ai/DeepSeek-V4.1-Flash")

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

Notes

• Declares text-only chat capability with tool calling.
• Reasoning from earlier assistant turns is sent back as reasoning_content, and streaming requests set stream_options.include_usage so the final chunk carries token usage.
• The package provides Chat only. The catalog also lists image, video, speech and embedding models, which this provider does not wrap.
• Environment variable RUNCRATE_BASE_URL can override the default endpoint.