API Documentation

For AI Agents

You are an AI agent. Here is everything you need:

Base URL:

Text/Chat: POST /v1/chat/completions with "provider": "ollama" (fastest) or "provider": "opencode-go" (best QA)

Vision QA: POST /v1/chat/completions with "provider": "opencode-go", "model": "kimi-k2.5", "max_tokens": 200000. Send image as base64 in content array.

Image Gen: POST /v1/images/generations with "provider": "arli"

Models: GET /v1/models?provider=opencode-go|ollama|featherless|arli|arli-image

Analytics: GET /v1/analytics — usage stats, model performance, call history (auto-logged)

Model Recommendations

Task	Model	Provider	Why
Visual QA / Screenshot Review	`kimi-k2.5`	`opencode-go`	200K tokens, best analysis quality, handles thinking
Visual QA (fast)	`kimi-k2.5`	`ollama`	2.4x faster, same model, /api/generate endpoint
Fastest Text	`gemma3:4b`	`ollama`	0.9s response time
Best Quality Text	`gpt-oss:120b`	`ollama`	111.8 t/s throughput, 2.2s
Best Reasoning	`deepseek-v3.2`	`ollama`	688B params, 5.9s
Image Generation	`FLUX.2-klein-4B`	`arli`	16.6s, high quality
Image Editing	`FLUX.2-klein-4B`	`arli`	Best quality edits
Image Upscaling	`4x-UltraSharp`	`arli`	4x upscaling
Video Generation	`Wan 2.1 1.3B`	`hf-inference`	Free, fast video generation
Coding	`glm-5`	`opencode-go`	Optimized for code
Fallback Text	`Llama-3.3-70B-Instruct`	`arli`	6.7s, reliable

Endpoints

Method	Path	Description
POST	`/v1/chat/completions`	Chat/text/vision generation (all providers)
POST	`/v1/completions`	Text completion
POST	`/v1/images/generations`	Image generation (Arli)
POST	`/v1/images/edits`	Image editing (Arli)
POST	`/v1/images/upscale`	Image upscaling (Arli)
GET	`/v1/models`	List all models (filter by `?provider=`)
POST	`/v1/tokenize`	Token counting (Featherless)
GET	`/v1/health`	Health check
GET	`/v1/status`	Live provider status & model counts
GET	`/v1/routes`	All available endpoints
GET	`/v1/recommendations`	Best model per task (filter `?category=`)
GET	`/v1/analytics`	API usage analytics and model performance dashboard

Examples

Visual QA Review (best: Kimi K2.5 via OpenCode Go)

Send a screenshot, get a detailed QA report with issues and fixes.

curl -X POST /v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k2.5",
    "provider": "opencode-go",
    "max_tokens": 4000,
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "image_url", "image_url": {"url": "data:image/png;base64,YOUR_IMAGE_BASE64"}},
          {"type": "text", "text": "Perform a visual QA review. List Critical/Moderate/Minor issues with fixes."}
        ]
      }
    ]
  }'

Vision QA via Ollama (faster)

curl -X POST /v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k2.5",
    "provider": "ollama",
    "max_tokens": 4000,
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "image_url", "image_url": {"url": "data:image/png;base64,YOUR_IMAGE_BASE64"}},
          {"type": "text", "text": "Describe this image."}
        ]
      }
    ]
  }'

Generate an Image

curl -X POST /v1/images/generations \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A serene mountain landscape at sunset",
    "model": "FLUX.2-klein-4B",
    "n": 1,
    "size": "1024x1024"
  }'

Quick Text Generation

curl -X POST /v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemma3:4b",
    "provider": "ollama",
    "messages": [{"role": "user", "content": "Explain quantum computing in 2 sentences."}],
    "max_tokens": 100
  }'

List Models by Provider

# All providers
curl /v1/models

# OpenCode Go only (Kimi K2.5, GLM-5, MiniMax)
curl /v1/models?provider=opencode-go

# Ollama Cloud (34 models)
curl /v1/models?provider=ollama

# Arli AI image models
curl /v1/models?provider=arli-image

Check System Status

curl /v1/status | python3 -m json.tool

Providers

OpenCode Go	Online	Kimi K2.5, GLM-5, MiniMax M2.5/M2.7	`?provider=opencode-go`
Ollama Cloud	Online	34 models, vision + reasoning, 3 concurrent	`?provider=ollama`
Featherless AI	Online	5,757 text models, 15 categories	`?provider=featherless` (default)
Arli AI	Online	117 text + 80 image models	`?provider=arli`