Summer Engine MCP guide
How to make games for free with local LLMs in Summer Engine
Run a free local model — Ollama, LM Studio, or Goose — and build real games in Summer Engine through MCP. No subscription; your GPU does the thinking. Summer Engine is the desktop game engine; Local LLMs drives it through MCP with a model running on your own hardware.
Summer is free for MCP use
Download the desktop engine and connect it to your agent. The model runs on your own hardware — no AI subscription, no per-message cost.
Open Local LLMs
Run a free local model on your own GPU. The setup writes the MCP config so {agentLabel} can drive the engine directly.
Launch a real game
The local model writes code. Summer Engine shows the game, runs Play mode, and exports projects for Steam, desktop, mobile, and console pipelines.
Local setup notes
The 2026 specifics.
Easiest: LM Studio
One app for the model and the MCP host. The setup command above writes the config; pick a model, raise its context to 32k+, and build. See the LM Studio guide for details.
Terminal: OpenCode + Ollama
The most common 2026 pairing. Ollama serves the model, OpenCode is the agent. Start Ollama with OLLAMA_CONTEXT_LENGTH=65536 — the 4k default silently breaks MCP. See the Ollama guide.
What to run
gpt-oss-20b on 12–16 GB of VRAM, Qwen3-Coder-30B or Gemma 4 27B on 24 GB, Q4_K_M quantization or better. Engine control and asset search are free; cloud generation uses credits.
Why this matters
Local LLMs runs free on your hardware. Summer Engine shows the game.
A capable local model can operate a real engine through MCP: open scenes, place objects, run the project, and read errors. Your GPU does the thinking — no subscription, no per-token bill, and your prompts never leave your machine.