neural
engine
prompt diff · v2
two prompts walk into a bar — let's see what they cost.
◐
light
share
↗
add api key
run both
⌘↵
A.
via
Claude Sonnet 4.6
▾
75
input tokens
system
You are a senior technical writer. Explain concepts with precision, brevity, and concrete examples. Prefer lists when appropriate. Never hedge.
prompt
Explain vector embeddings to a backend engineer who knows databases but not ML. Use at most 3 paragraphs and include one SQL-style analogy.
cost, est.
$0.0014
per run
output tokens, est.
80
tok
A. response
edit prompt →
Claude Sonnet 4.6 says:
the press
?
▶
Press
⌘
↵
to run both
7
-token delta detected.
Ready when you are.
A ledger will appear here once both sides have replied.
B.
via
Claude Sonnet 4.6
▾
82
input tokens
system
You are a senior technical writer. Explain concepts with precision, brevity, and concrete examples. Prefer lists when appropriate. Never hedge.
prompt
+14
−4
vs A
Explain vector embeddings to a backend engineer who knows databases but not ML. Use at most 3 paragraphs, include one SQL-style analogy, and end with a one-line tl;dr.
Explain
vector
embeddings
to
a
backend
engineer
who
knows
databases
but
not
ML.
Use
at
most
3
paragraphs,
include
one
SQL-style
analogy,
and
end
with
a
one-line
tl;dr.
cost, est.
$0.0014
per run
output tokens, est.
80
tok
B. response
edit prompt →
Claude Sonnet 4.6 says: