Llama

49.2 experience score ▲ 2.0 Confidence: High

31 opinions tracked this week · 7-day rolling window

30-day experience score

Version breakdown

Only comments that explicitly name a version contribute to its score.

CodeLLaMa 7B Instruct Solidity	mentions only		n=1	Confidence: Insufficient
Llama 3.3 70B Instruct	mentions only		n=1	Confidence: Insufficient

30 generic family mentions are included in the family score, not any version score.

By experience dimension

General text	46	n=8
Coding	not enough data	n=2
Reasoning	not enough data	n=2
Image / vision	not enough data	n=0
Roleplay / creative	not enough data	n=0
Safety & refusals	not enough data	n=0
Speed & latency	48	n=16
Local deploy	not enough data	n=3

Top positive drivers

speed quality Speed & latency ×5
tool failure Coding ×1
customization ease Local deploy ×1
low end hardware Local deploy ×1
memory flexibility Local deploy ×1

Top negative drivers

performance issue Speed & latency ×4
speed quality Speed & latency ×2
performance issue General text ×2
tool failure Coding ×1
model quality Local deploy ×1

What users are saying

Selected from the latest 7 days, prioritizing recency, evidence quality, confidence, and engagement.

🟢 “undertrained but easy to finetune”
via reddit · local_deploy · en · 2026-07-17

🟢 “llama3.3: 70b is quite good model tbh”
via reddit · text_general · en · 2026-07-17

🔴 “It's a custom 3 bit quant of Llama 3.1 8B and other shortcuts. The quant is not good.”
via hackernews · text_general · en · 2026-07-17

🔴 “If anything, Kimi K3 feels easier than running Llama3-70B or 405B... I love how you say "Yet we did" as if any significant percentage of us were able to run those models lmao”
via reddit · speed_latency · en · 2026-07-17