Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.magnitude.dev/llms.txt

Use this file to discover all available pages before exploring further.

The coding plan is $20/mo. 3 day free trial, no card required. Add a payment method to continue with the plan after trial.

What’s included

  • Best models, always. We benchmark constantly. New model wins, you get it.
  • Verified providers. Only the ones serving the models correctly and fast.
  • Zero data retention. Our providers do not train on your code.
  • Built-in web search. Exa is wired in by default.
  • Generous rate limits. Sized for full days of agent work. Limits reset every 5 hours.

Models in the lineup

The current lineup of models powering the agent team:
  • Kimi K2.6
  • DeepSeek V4 Pro
  • GLM 5.1
  • MiniMax M2.7
Each model is assigned to the agent it performs best for. See Agent Team for the full mapping. We test these constantly. New models drop, the lineup updates.

Usage estimates

Rate limits reset every 5 hours. Here’s roughly what you can expect per model based on our testing.

Requests per window

Model5 hoursWeeklyMonthly
DeepSeek V4 Pro1,8324,5819,162
GLM 5.11,1452,8635,727
Kimi K2.61,8074,5189,037
MiniMax M2.74,93812,34524,691

Tokens per window

Model5 hoursWeeklyMonthly
DeepSeek V4 Pro112M279M558M
GLM 5.170M175M349M
Kimi K2.6111M276M551M
MiniMax M2.7301M752M1.6B
Estimates assume a typical request of 700 input tokens, 60,000 cached input tokens, and 200 output tokens. Actual usage will vary based on task complexity.