Head-to-head comparison across 2benchmark categories. Overall scores shown here use BenchLM's provisional ranking lane.
Composer 2
73
Laguna XS.2
32
Pick Composer 2 if you want the stronger benchmark profile. Laguna XS.2 only becomes the better choice if you want the cheaper token bill.
Agentic
+31.6 difference
Coding
+4.7 difference
Composer 2
Laguna XS.2
$0.5 / $2.5
$0 / $0
N/A
N/A
N/A
N/A
200K
131K
Pick Composer 2 if you want the stronger benchmark profile. Laguna XS.2 only becomes the better choice if you want the cheaper token bill.
Composer 2 is clearly ahead on the provisional aggregate, 73 to 32. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Composer 2's sharpest advantage is in agentic, where it averages 61.7 against 30.1. The single biggest benchmark swing on the page is Terminal-Bench 2.0, 61.7% to 30.1%.
Composer 2 is also the more expensive model on tokens at $0.50 input / $2.50 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Laguna XS.2. That is roughly Infinityx on output cost alone. Composer 2 gives you the larger context window at 200K, compared with 131K for Laguna XS.2.
Composer 2 is ahead on BenchLM's provisional leaderboard, 73 to 32. The biggest single separator in this matchup is Terminal-Bench 2.0, where the scores are 61.7% and 30.1%.
Composer 2 has the edge for coding in this comparison, averaging 58 versus 53.3. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
Composer 2 has the edge for agentic tasks in this comparison, averaging 61.7 versus 30.1. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.
Free. No spam. Unsubscribe anytime.