Credits, tokens and AI models

Codexe does not charge per project. It charges for the work the model actually does, measured in tokens - the units language models read and write. A short prompt that produces a small plugin costs a fraction of a large multi-file mod.

Both directions count: the prompt and surrounding project context you send in, and the code the model writes back. Output is the expensive side, which is why iterating on one file is much cheaper than regenerating a whole project.

The free tier refills a 50K token allowance daily, so you can keep experimenting without a subscription. Paid plans raise the monthly allowance and unlock more capable models.

Current pricing and the full feature comparison live on the pricing page. Teams that need priority generation and the largest models can read about the Business tier.

Codexe is not tied to one AI vendor. The catalogue spans models from Anthropic, OpenAI, DeepSeek and xAI, and each model is gated to a minimum plan tier based on what it costs to run.

• Free tier - fast, low-cost models suited to small plugins, scripts and edits.
• Pro - adds frontier reasoning models and long-context variants for bigger projects.
• Business and above - adds the largest models in the catalogue.

The model picker in the workspace shows exactly which models your current plan can select, so the list you see is always accurate for your account.

Paid plans can also set how much deliberation a model does before answering. Higher effort spends more reasoning tokens and generally produces better results on complex, multi-file work; lower effort is cheaper and faster for small edits. Free accounts run at the medium default.

When your balance is too low for a request, generation stops rather than producing a truncated project. Free accounts get their allowance back on the next daily refill; paid accounts can top up with a token pack or move to a higher tier.