Me
llama-4-scout-17b-16e-instruct Benchmark & Insights
Meta Cloudflare Workers AI
Updated Apr 13, 2026 All models
Sample size
19 runs
in window
Accuracy
63.2%
consensus match · 19d
Confidence
88%
over 19 runs
Window end
Apr 13, 2026
most recent run
Input price
$0.27/MTok
prompt tokens
Output price
$0.85/MTok
completion tokens
Model insights
- 01 Diverged on 7 of its 19 days, with 5 false "unsafe" flips packed into a late-March/early-April streak.
- 02 A clear regression from the older llama-3.3-70b sibling, which scores 20 points better over a far longer record.
Recent forecasts