The editor is free. Forever.

Plug in your own AI key and pay nothing. Or use ours: buy credits, spend them per request, top up when you want. No subscription, no seats, nothing expires.

Mathend
Free
$0forever
  • Unlimited derivations and notes
  • PDF and DOCX import and export
  • Notes stay on your machine — no account needed
  • Bring your own key: Claude, GPT, Gemini, Copilot, and more
  • All future updates
Agent credits
Prepaid

No API key to manage. Every request is billed at cost plus a small margin, deducted from your balance.

$5
Try the hosted agent.
$10
A month of steady use.
$25
Heavy agent days.

Credits never expire. Unused balance stays on your account.

What credits buy

Fourteen models in three tiers. You are charged per token, so a short edit costs a fraction of a cent and only long sessions add up.

ModelInput / 1MOutput / 1M
EconomyEdits, formatting, bulk rewrites — most turns cost a fraction of a cent.
Hy3
The cheapest thing that still calls tools reliably.
$0.343$1.37
DeepSeek V4 Flash
Very cheap per token and quick to answer. Good default for drafts.
$0.364$0.728
MiMo-V2.5
Cheap, long context, steady on mechanical edits.
$0.364$0.728
MiniMax M3
Fast and cheap. Good for edits, formatting, and bulk rewrites.
$0.780$3.12
Qwen3.7 Plus
Balanced budget pick with a full million tokens of context.
$0.832$3.33
DeepSeek V4 Pro
Stronger reasoning than Flash, still inside the economy band.
$1.13$2.26
MiMo-V2.5-Pro
The larger MiMo. Cheap output for how much it gets right.
$1.13$2.26
StandardMulti-step document work that has to come out right the first time.
GLM-5.2
Reliable tool-caller. A good step up when economy models stall.
$1.34$4.21
Kimi K2.7 Code
Tuned for code and structured output. 262k context.
$1.50$7.00
Qwen3.7 Max
The big Qwen. Handles long documents without losing the thread.
$2.95$8.85
Grok 4.5
Fast for its size, and cheap on output. 500k context.
$4.00$12.00
Claude Sonnet 5
The everyday default. Strong at multi-step document work.
$4.00$20.00
PremiumHard derivations and long proofs, where a wrong answer costs more than the tokens.
Kimi K3
Frontier-class reasoning at a lower rate than Opus.
$4.80$24.00
Claude Opus 5
Hardest derivations and long proofs. Slower, and worth it.
$8.00$40.00

Rates can change when upstream prices do. The model menu in the app always shows the current one.