The coders

Six AI models can be your coder. Luka rates each one for modding. The workbench shows the coders your account can use right now and what each one costs before you send a message.

LR
Luka's Rating, out of 10: how Luka, who built the coder, rates each model overall for modding. What you get for what it costs.
Power, out of 10: Luka's rating of the model's own intelligence when it makes mods, with the price left out.
Anthropic

Claude Opus 5.5

claude-opus-5-5

“Luka's preferred current model.”

Luka
Google

Gemini 4

gemini-4-argon

“As strong as GPT-6 Astra, half the price of Opus, and twice as fast. Do I need to say more?”

Luka
OpenAI

GPT-6 Astra

gpt-6-astra

“The best in the GPT family, but still too expensive. Opus is better and cheaper, in my opinion.”

Luka
OpenAI

GPT-6.1 Sol

gpt-6.1-sol

“Cheaper than Astra and slightly worse. Use it if its price ever drops below Opus.”

Luka
Google

Gemini 3.8 Flash

gemini-3.8-flash

“Super cheap and the fastest of them all, but not really bright for complex tasks.”

Luka
Anthropic

Claude Fable 5.1

claude-fable-5-1

“About as strong as Opus, but needlessly expensive.”

Luka

Prices and groups

Every model runs at Astropond's prices. VibeModder adds nothing on top, and the modding tools themselves are free.

Automatic

Each message goes to the cheapest healthy Astropond group serving the model. The price can change when a group goes down.

Pinned

Pin a group in the coder picker for a price that stays put. Unpin it to go back to automatic.

Live prices are in the app

Prices per group move with Astropond's live rates, so this page does not list them. The workbench's coder picker shows the price you would pay right now, for every coder and group.

Try a coder on your mod

Open the workbench, pick a game and a coder, and describe the change you want. Your Astropond login and balance work here.