Mistral

48.3 As usual experience score ▼ 2.7 Confidence: High

37 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.

Mistral Small 4	43.5−4.8	n=5	Confidence: Low
Mistral Large	mentions only	n=2	Confidence: Insufficient
Mistral Nemo	mentions only	n=1	Confidence: Insufficient
Mistral Small 3	mentions only	n=1	Confidence: Insufficient
Mistral Small 3.1 24B	mentions only	n=1	Confidence: Insufficient

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

By experience dimension

General text	41	n=17
Coding	not enough data	n=1
Reasoning	not enough data	n=3
Image / vision	not enough data	n=0
Roleplay / creative	not enough data	n=4
Safety & refusals	not enough data	n=0
Speed & latency	45	n=11
Local deploy	not enough data	n=1

Top positive drivers

creative surprise Roleplay / creative ×3
speed quality Speed & latency ×2
competitor flight General text ×2
coding preference Coding ×1
strong reasoning Reasoning ×1

Top negative drivers

competitor flight General text ×4
performance issue Speed & latency ×3
looping issue Reasoning ×1
quality down Reasoning ×1
kv cache vram overflow Speed & latency ×1

What users are saying

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

🟢 “Mistral Small 24B's finetuning still excels in creative tasks.”
via reddit · roleplay_creative · en · 2026-07-21

🔴 “quality wise, they still were worse than chinese or us models”
via reddit · text_general · en · 2026-07-21

⚪ “relatively fast, not a great model overall”
via reddit · speed_latency · en · 2026-07-21

🟢 “I love the Crimson-Constellation-12B (Mistral Nemo merge) and Cydonia-24B-V4.3 in RP - despite the advent of Gemma 4 these models have their own distinct style.”
via reddit · roleplay_creative · en · 2026-07-21