Head-to-head comparison across 1benchmark categories. Overall scores shown here use BenchLM's provisional ranking lane.
Command A+
32
Qwen3.5 397B
64
Verified leaderboard positions: Command A+ unranked · Qwen3.5 397B #17
Pick Qwen3.5 397B if you want the stronger benchmark profile. Command A+ only becomes the better choice if you want the stronger reasoning-first profile.
Multimodal
+19.8 difference
Command A+
Qwen3.5 397B
$2.5 / $10
$0.6 / $3.6
272 t/s
96 t/s
0.25s
2.44s
128K
128K
Pick Qwen3.5 397B if you want the stronger benchmark profile. Command A+ only becomes the better choice if you want the stronger reasoning-first profile.
Qwen3.5 397B is clearly ahead on the provisional aggregate, 64 to 32. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Qwen3.5 397B's sharpest advantage is in multimodal & grounded, where it averages 79.6 against 59.8. The single biggest benchmark swing on the page is CharXiv, 52.7% to 80.8%.
Command A+ is also the more expensive model on tokens at $2.50 input / $10.00 output per 1M tokens, versus $0.60 input / $3.60 output per 1M tokens for Qwen3.5 397B. That is roughly 2.8x on output cost alone. Command A+ is the reasoning model in the pair, while Qwen3.5 397B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use.
Qwen3.5 397B is ahead on BenchLM's provisional leaderboard, 64 to 32. The biggest single separator in this matchup is CharXiv, where the scores are 52.7% and 80.8%.
Qwen3.5 397B has the edge for multimodal and grounded tasks in this comparison, averaging 79.6 versus 59.8. Inside this category, CharXiv 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.