Qwen3.8 Flash · Model reviews

PositiveLocal deploy2026-09-04

“Damn son”

NegativeGeneral text2026-09-04

“i did notice it over engineers as well...”

PositiveCoding2026-09-04

“the final result was perfect: no bugs, and everything magically implemented and working. The time investment is worth it with this model”

NegativeReasoning2026-09-04

“hallucinates in the middle”

NegativeReasoning2026-09-04

“It's fine for pure execution, but lacking in planning. The intuition for handling business problems isn't great, not smart enough.”

ZhihuzhMachine translated

Show original

做纯粹的执行问题不大，做规划就比较欠缺了，对业务问题的处理直觉不太好，不够聪明

PositiveLocal deploy2026-09-03

“enabling cache makes it get about 20 tokens/s with a r9700 / rx9070 with 256k context / kv on f16 & 128gb ddr4”

PositiveGeneral text2026-09-03

“It is a bit smaller while losing hardly any precision”

PositiveReasoning2026-09-03

“I really love this model so far”

NegativeSpeed & latency2026-09-02

“The slowness is unbearable, I won't use it anymore.”

ZhihuzhMachine translated

Show original

慢的忍不了，不会用它了

NegativeLocal deploy2026-09-02

“Qwen3.8 flash is like an idiot, either stops working halfway or has various hallucinations and tool call failures.”

ZhihuzhMachine translated

Show original

Qwen3.8 flash犹如一个智障，要么运行一半不干活了，要么各种幻觉工具调用失败

NegativeLocal deploy2026-09-01

“Running this model seems to require 128g memory plus 16g vram. Now 128g memory also costs over 10,000 yuan.”

ZhihuzhMachine translated

Show original

运行这个模型，好像要128g内存加16g显存现在128g内存也要1万多

NegativeGeneral text2026-09-01

“thinking default is x-high super annoying and was even hallucinating/looping quite often”

MixedCoding2026-08-31

“qwen3.8-flash >＝ dsv4f > glm-5.3-flash, but Qwen really does go into full-on 'exam-taking' mode”

ZhihuzhMachine translated

Show original

qwen3.8-flash >＝ dsv4f > glm-5.3-flash 但qwen确实会雷霆大考

Community reviews