AI, Learned Today
Search...
⌘K
Github
Substack
Open main menu
Logs
2026-05-14
2026-05-14
2026-05-14
Added
stephen-reid-coding-agents-benchmarks
- Coding agent benchmarks framed as cost vs speed: once models are “good enough”, tasks/
a
n
d
t
i
m
e
/
t
a
s
k
m
a
t
t
e
r
;
S
u
b
s
t
a
c
k
n
o
t
e
s
C
u
r
s
o
r
C
o
m
p
o
s
e
r
2
a
s
a
m
a
j
o
r
o
u
t
l
i
e
r
(
14
t
a
s
k
s
/
and time/task matter; Substack notes Cursor Composer 2 as a major outlier (~14 tasks/
an
d
t
im
e
/
t
a
s
kma
tt
er
;
S
u
b
s
t
a
c
kn
o
t
es
C
u
rsor
C
o
m
p
oser
2
a
s
amaj
oro
u
tl
i
er
(
14
t
a
s
k
s
/
at ~521s).
Edit this page