Head-to-head comparison across 1benchmark categories. Overall scores shown here use BenchLM's provisional ranking lane.
GPT-5.4 mini
70
Step 3.7 Flash
72
Pick Step 3.7 Flash if you want the stronger benchmark profile. GPT-5.4 mini only becomes the better choice if you need the larger 400K context window.
Agentic
+0.3 difference
GPT-5.4 mini
Step 3.7 Flash
$0.75 / $4.5
$0.2 / $1.15
201 t/s
N/A
3.85s
N/A
400K
256K
Pick Step 3.7 Flash if you want the stronger benchmark profile. GPT-5.4 mini only becomes the better choice if you need the larger 400K context window.
Step 3.7 Flash has the cleaner provisional overall profile here, landing at 72 versus 70. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
Step 3.7 Flash's sharpest advantage is in agentic, where it averages 65.9 against 65.6. The single biggest benchmark swing on the page is Terminal-Bench 2.0, 60% to 59.5%.
GPT-5.4 mini is also the more expensive model on tokens at $0.75 input / $4.50 output per 1M tokens, versus $0.20 input / $1.15 output per 1M tokens for Step 3.7 Flash. That is roughly 3.9x on output cost alone. GPT-5.4 mini gives you the larger context window at 400K, compared with 256K for Step 3.7 Flash.
Step 3.7 Flash is ahead on BenchLM's provisional leaderboard, 72 to 70. The biggest single separator in this matchup is Terminal-Bench 2.0, where the scores are 60% and 59.5%.
Step 3.7 Flash has the edge for agentic tasks in this comparison, averaging 65.9 versus 65.6. Inside this category, Toolathlon 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.