Qwen
1,223 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.
| Qwen3.8 27B | 63.2+0.9 | n=458 | Confidence: High | |
|---|---|---|---|---|
| Qwen3.6 35B A3B | 64.4+2.1 | n=35 | Confidence: High | |
| Qwen3.6 27B | 53.8−8.5 | n=21 | Confidence: Medium | |
| Qwen3.8 Max | 53.5−8.8 | n=7 | Confidence: Low | |
| Qwen3.5-122B-A10B | mentions only | n=4 | Confidence: not enough data |
710 generic family mentions are included in the family score, not any version score.
By experience dimension
| General text | 53.0 | n=257 | |
|---|---|---|---|
| Coding | 76.6 | n=190 | |
| Reasoning | 73.1 | n=209 | |
| Image / vision | 71.7 | n=21 | |
| Video generation | not enough data | n=0 | |
| Roleplay / creative | 56.0 | n=12 | |
| Safety & refusals | 38.9 | n=10 | |
| Speed & latency | 54.1 | n=411 | |
| Local deploy | 62.3 | n=130 |
Top positive drivers
- speed quality Speed & latency ×139
- strong reasoning Reasoning ×113
- problem solved Coding ×29
- problem solved General text ×27
- competitor flight General text ×22
Top negative drivers
- performance issue Speed & latency ×67
- speed quality Speed & latency ×38
- performance issue General text ×23
- competitor flight General text ×18
- performance issue Local deploy ×16
What users are saying
Selected from the latest 7 days, prioritizing recency, evidence quality, confidence, and engagement.
⚪ “Let me do a RAG testing, even I know that qwen will do better there. But I'm not running it on xhigh ever again, 30hrs of gpu time, against 3-4 hrs at medium”
via Reddit · Speed & latency · 2026-08-26
🟢 “oh yes. very good at vision”
via Reddit · Image / vision · 2026-08-26
🟢 “Qwen 27B starts to shine when you give it some kind of RAG. It's not a memory castle, it's a tinkering machine. You give it a folder to a library that it was not trained on, you ask it "how do i do ... that...?" and it figures it out.”
via Reddit · General text · 2026-08-26
🟢 “they have really taught the machine to think, or at the very least rationally and logically follow chains of evidence until answer is obtained”
via Reddit · Reasoning · 2026-08-26
🟢 “Qwen always had the best vision for local in my own tests so I'm not surprised”
via Reddit · Image / vision · 2026-08-26