Current AI Benchmarks (Week 6, 2026)
Key Takeaways (Week 6):
The Value Leader: Liquid AI sweeps the top 2 spots. Their LFM2 models are ~50% cheaper than the competition, giving them the highest Efficiency Scores despite moderate latency.
The Speed Demons: If latency is your priority, Ministral 3B (#5) and Llama Guard 3 8B (#4) are the clear winners, both clocking in under 0.20s.
Small is Big: The entire Top 5 is dominated by efficient models under 10B parameters. The era of massive, expensive models for everyday tasks is ending.
🏆 Top 5 Models (Ranked by Efficiency):
Rank
Model Name
Provider
Cost (1M Tokens)
Latency (s)
Score
#1
AI LFM2 2.6B
Liquid
$0.0125
0.47s
14,134
#2
AI LFM2 8B
Liquid
$0.0125
0.55s
12,345
#3
Llama 3.2 3B Instruct
Meta
$0.0200
0.32s
11,961
#4
Llama Guard 3 8B
Meta
$0.0300
0.18s
11,737
#5
Ministral 3B
Mistral
$0.0400
0.17s
9,293
Updated every Monday at 8 AM EST. Want the raw CSV?