This video segment addresses the high cost of using Claude AI, specifically mentioning that it costs $75 per million tokens and that a $20 subscription offers no benefits regarding credits. It then proposes three solutions to mitigate these costs.

1.  **Caveman Prompting:** The first solution involves using a "Caveman" prompt, which is credited with achieving 30,000 stars. This prompt style is designed to reduce the "BS" (unnecessary output) by 75%, focusing only on important information to save on expensive output tokens.
2.  **Context Clearing:** The second recommendation is to frequently use the command `{slash}clear` within Claude's code interface. This action is crucial for reducing the input context window, thereby lowering token usage.
3.  **Utilizing Cheaper/Free Models:** The third strategy suggests using more affordable or entirely free AI models when feasible. The video highlights Minimax 2.7 as a free model available on Nvidia, which can be integrated with open-source tools like "open Claude" and other "agentic terminals." The speaker offers to send a full guide to those who comment "Claude" on the video.

Singh in USA

This is the answer for everyone requesting free Claude credits and wasting tokens because Claude charges $75 per million tokens and $20 subscription of Claude will give you nothing.

But now we have a solution.

Number one, caveman. This guy actually got 30,000 stars for just writing a prompt, which actually cut downs the BS of output because output tokens are expensive. So it reduces the output by 75%. It only releases what is important.

Number two, you really got a clear context by {slash} clear on Claude code as often as possible because it will reduce the input context window.

And number three, use cheaper models and free models whenever possible. This model, Minimax 2.7, is absolutely free on Nvidia, which you can use right now and connect with open Claude and connect with any open source agentic terminal.

So comment down Claude and I'll send you the full guide in your DM.

Get Claude Code for free & cut 75% costs!!

This is the answer for everyone
requesting free Claude credits and
wasting tokens because Claude charges
$75 per million tokens and $20
subscription of Claude will give you
nothing. But now we have a solution.
Number one, caveman. This guy actually
got 30,000 stars for just writing a
prompt, which actually cut downs the BS
of output because output tokens are
expensive. So it reduces the output by
75%. It only releases what is important.
Number two, you really got a clear
context by {slash} clear on Claude code
as often as possible because it will
reduce the input context window. And
number three, use cheaper models and
free models whenever possible. This
model, Minimax 2.7, is absolutely free
on Nvidia, which you can use right now
and connect with open Claude and connect
with any open source agentic terminal.
So comment down Claude and I'll send you
the full guide in your DM.

Get Our Mobile App

Get Claude Code for free & cut 75% costs!!

Transcription