Qwen 3.8 Flash
Hacker ApprovedRanked 34 of 72 language models by recommenders
People recommend it for coding, agentic workflows and tool calling, and call it smart enough, fast and very cheap. Some mixed reports say outputs are sometimes inconsistent.
People recommend it for coding, agentic workflows and tool calling, and call it smart enough, fast and very cheap. Some mixed reports say outputs are sometimes inconsistent.
23 people gave an opinion
- recommend it
- 23
- criticize it
- 0
- approval share
- 100%
Recommended in 6 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: not recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: recommended (month in progress)
Why it carries the badge
- 23 people recommend itNeeds 10 or more, met.
- Nobody criticizes itNeeds 2 to 1 or better, met.
- Recommended in 6 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Fast with long context windows
“It thinks a little less and thinks a little faster, adding up to a real improvement.”
Comment from 25 Oct 2026 in “Five Qwen3.8-27B models match Fable 5 Using GVS5H Harness” “can run a dozen or so sessions of qwen 3.8 flash next at once at > 100tps each”
Comment from 30 Oct 2026 in “Anthropic Cracking Down on Video Game Rewrites” “it's doing great, really fast and super long context windows”
Comment from 2 Oct 2026 in “From the creator of Redis; run LLM locally with ds4” Smart enough and very cheap
“smart enough, tool calling works really well, and very cheap”
Comment from 29 Apr 2026 in “He asked AI to count carbs 27000 times. It couldn't give the same answer twice” “tool calling works well, it's smart enough, and very cheap”
Comment from 29 Apr 2026 in “He asked AI to count carbs 27000 times. It couldn't give the same answer twice” “it was nice going to a model that is "smart enough"”
Comment from 22 Sep 2026 in “The Economics of Open-Weight Inference” Strong for coding tasks
“That is currently the best local model for actual coding tasks in my + many coworkers experience.”
Comment from 20 Oct 2026 in “MiniMax M3.1 Flash” “Qwen3.8-Flash is good enough (Opus 4.7 level) for all of my coding needs.”
Comment from 7 Oct 2026 in “What is Codemode” Works well for agentic tool calling
“Seems like in agentic work flow the qween flash and Deepseek flash models are quite good.”
Comment from 3 May 2026 in “Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge” “Overall this model is very good on long horizon agentic work.”
Comment from 21 Sep 2026 in “M5 Ultra Mac Studio Review”
What people warn about
Each mark is one comment
Outputs sometimes inconsistent
“I'm using Eugr's recipe (TP=1) with Nvidia's NVFP4 quant and I'm still seeing it sometimes!”
Comment from 18 Sep 2026 in “Qwen 3.8 Omni Flash” Quantized variant seems degraded
“You are better off going with a lobotomized 3.8 flash MoE for the same amount of memory.”
Comment from 21 Oct 2026 in “Qwen3-Coder writes less secure code when told it's for a US persona”
Other language models people recommend
The four most recommended other language models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| Claude Opus | 3,331 | 2,746 | 55% | 13 | 4 mo156 owners | — |
| OpenAI GPT | 1,205 | 987 | 55% | 13 | 6 mo47 owners | — |
| Claude Sonnet | 854 | 586 | 59% | 13 | 6 mo46 owners | — |
| DeepSeek V4 Flash | 828 | 223 | 79% | 8 | 1 mo24 owners | Approved |
| Qwen 3.8 FlashThis page | 23 | 0 | 100% | 6 | — | Approved |
What people said
26 of the 27 comments that mention it, newest first. Each links to the original comment.
“Really excited for Qwen4-Flash when it comes out.”
“can run a dozen or so sessions of qwen 3.8 flash next at once at > 100tps each”
“It thinks a little less and thinks a little faster, adding up to a real improvement.”
“You are better off going with a lobotomized 3.8 flash MoE for the same amount of memory.”
“I'm very happy with the outputs one of the Flash models has been generating.”
“That is currently the best local model for actual coding tasks in my + many coworkers experience.”
“my understanding is that flash next can be very efficient wrt ram needed”
“Qwen3.8 flash next a close second”
“Qwen3.8-Flash is good enough (Opus 4.7 level) for all of my coding needs.”
“Qwen 3.8 flash has also good reviews from home labbers.”
“it is at least as good as Sonnet 5 in benchmarks like DeepSWE”
“Qwen 3.8 Flash locally on my DGX Spark-like”
“it's doing great, really fast and super long context windows”
“it was nice going to a model that is "smart enough"”
“Going from Opus 5 to Qwen3.8 flash has been a breath of fresh air.”
“Overall this model is very good on long horizon agentic work.”
“I'm using Eugr's recipe (TP=1) with Nvidia's NVFP4 quant and I'm still seeing it sometimes!”
“Can confirm here as well.”
“is so good to not slog through the regular loop of re”
“produced correct results... matched my own hand calculations perfectly”
“Qwen 3.8 Flash has 6B active and it smokes both models.”
“it does a good job for my fiddling around”
“the hit should be below 10% performance”
“Seems like in agentic work flow the qween flash and Deepseek flash models are quite good.”
“smart enough, tool calling works really well, and very cheap”
“tool calling works well, it's smart enough, and very cheap”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.