# 🦢 Goose Custom Providers Setup

Three custom providers configured for FREE and low-cost models!

---

## 🆓 OpenCode (opencode_free)

**Base URL:** `https://opencode.ai/zen/v1/chat/completions`

### Free Models:
| Model | Name | Context |
|-------|------|---------|
| `big-pickle` | **Big Pickle** (FREE!) | 128K |
| `minimax-m2.5-free` | MiniMax M2.5 Free | 128K |
| `gpt-5-nano` | GPT 5 Nano (FREE!) | 128K |

### Paid Models (cheap):
| Model | Input | Output |
|-------|-------|--------|
| `glm-5` | $1.00/M | $3.20/M |
| `glm-4.7` | $0.60/M | $2.20/M |
| `kimi-k2.5` | $0.60/M | $3.00/M |
| `claude-sonnet-4-6` | $3.00/M | $15.00/M |

---

## 🟢 NVIDIA NIM (nvidia_nim)

**Base URL:** `https://integrate.api.nvidia.com/v1/chat/completions`

### Free Models:
| Model | Name | Context |
|-------|------|---------|
| `meta/llama-3.3-70b-instruct` | **Llama 3.3 70B** | 128K |
| `meta/llama-4-scout-17b-16e-instruct` | **Llama 4 Scout** | 128K |
| `qwen/qwen3-32b` | **Qwen 3 32B** | 128K |
| `nvidia/llama-3.1-nemotron-70b-instruct` | **Nemotron 70B** | 128K |
| `microsoft/phi-4-multimodal-instruct` | **Phi 4 Multimodal** | 128K |

### Paid Models:
| Model | Input | Output |
|-------|-------|--------|
| `meta/llama-4-maverick-17b-128e-instruct` | $0.20/M | $0.50/M |
| `deepseek-ai/deepseek-r1` | $0.20/M | $0.50/M |

---

## ⚡ Kilo Gateway (kilo_gateway)

**Base URL:** `https://api.kilo.ai/v1/chat/completions`

### Free Models:
| Model | Name |
|-------|------|
| `kilo/auto-free` | **Auto Free** (Smart Routing) |
| `opencode/glm-5-free` | GLM 5 Free |
| `openrouter/qwen/qwen3-32b` | Qwen 3 32B |
| `openrouter/groq/llama-3.3-70b-versatile` | Llama 3.3 70B |

---

## 🚀 Quick Start

```bash
# Use the switcher
source goose-custom

# Examples:
source goose-custom opencode      # Big Pickle (FREE)
source goose-custom nvidia        # Llama 3.3 70B (FREE)
source goose-custom kilo          # Auto Free (FREE)

# Then start Goose
goose session
```

---

## 📁 Files Created

| File | Purpose |
|------|---------|
| `~/.config/goose/custom_providers/opencode_free.json` | OpenCode provider |
| `~/.config/goose/custom_providers/nvidia_nim.json` | NVIDIA NIM provider |
| `~/.config/goose/custom_providers/kilo_gateway.json` | Kilo Gateway provider |
| `~/.local/bin/goose-custom` | Switcher script |

---

## 🔑 API Keys Required

Add these to your environment:
```bash
export OPENCODE_API_KEY="your-opencode-key"
export NVIDIA_API_KEY="your-nvidia-key"
export KILOCODE_API_KEY="your-kilo-key"
```

---

## 💡 Recommendations

**For Coding:**
- `source goose-custom opencode` → Big Pickle (FREE, good for code)
- `source goose-custom nvidia-qwen` → Qwen 3 32B (FREE, great reasoning)

**For General Chat:**
- `source goose-custom nvidia` → Llama 3.3 70B (FREE, balanced)
- `source goose-custom kilo` → Auto Free (smart routing)

**For Long Context:**
- `source goose-custom opencode-kimi` → Kimi K2.5 (256K context)

---

*Setup completed: March 1, 2026*
*By: Krilly the Crab 🦀*