Qwen3-Omni-Flash
Omni model in AI tools
Recommended as a free-for-now omni model that works well and reports beating 2.5 Flash on benchmarks; critics say it struggles with non-speech sounds like laughter.
Recommended as a free-for-now omni model that works well and reports beating 2.5 Flash on benchmarks; critics say it struggles with non-speech sounds like laughter.
25 people gave an opinion
- recommend it
- 10
- criticize it
- 15
- approval share
- 40%
Recommended in 2 of 13 months
- November 2025: not recommended
- December 2025: recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: not recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 10 people recommend itNeeds 10 or more, met.
- 0.6 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 2 of 13 monthsNeeds 4 or more, not 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
Works well in practice
“the "the user didn't tell me to do anything" thoughts on OpenRouter”
Comment from 18 Sep 2026 in “Qwen 3.8 Omni Flash” “Yeah, it seemed pretty good.”
Comment from 18 Sep 2026 in “Qwen 3.8 Omni Flash” “Still works great though!”
Comment from 18 Sep 2026 in “Qwen 3.8 Omni Flash”
What people warn about
Each mark is one comment
Makes basic reasoning and language errors
“Weird, as someone not having a database of the web, I wouldn't be able to calculate either result.”
Comment from 10 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model” “I tried this out, and it's not passing the record (n.) vs. record (v.) test”
Comment from 11 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model” “It replied 29 resistors and provided a long list. Answer is 2 resistors”
Comment from 10 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model” - Comment from 10 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model”
Benchmark comparisons and blog questioned
“the image benchmarks are compared against 2.0, which seems suspicious”
Comment from 11 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model” “The blog post itself is technically unimpressive”
Comment from 18 Sep 2026 in “Qwen 3.8 Omni Flash” “This article feels a bit deceptive”
Comment from 10 Dec 2025 in “Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model”
What people said
27 of the 30 comments that mention it, newest first. Each links to the original comment.
“downloads automatically 441 MiB of mp4 files”
“Yeah, it seemed pretty good.”
“Still works great though!”
“The blog post itself is technically unimpressive”
“I see the same thing with Q8.”
“it appears to actually work pretty well, regardless of the occasional CoT amnesia”
“the "the user didn't tell me to do anything" thoughts on OpenRouter”
“That is a massive cost reduction.”
“Like, really weirdly compared to any model I've ever seen before.”
“it was ridiculously stupid”
“the image benchmarks are compared against 2.0, which seems suspicious”
“I tried this out, and it's not passing the record (n.) vs. record (v.) test”
“I've always been disappointed with Qwen Omni models in my English-first case”
“It still doesn't seem capable of producing any of the sounds, like laughter”
“Qwen3-Omni claims to support direct audio input into the model”
“clearly can't understand the difference between homophones”
“No cost to you for now, you just need a valid email to sign up on the platform.”
“I'm not convinced its end-to-end multimodal”
“was told my chat was blocked for inappropriate content”
“This article feels a bit deceptive”
“You basically can't use this model right now since none of the open source inference framework have the model fully implemented.”
“Looks to be API only. Bummer.”
“There aren't many open-weights omni models so I consider this a big deal.”
“crushing 2.5 Flash on every benchmark is huge”
“It's just not overly emotional. I definitely prefer it that way.”
“Weird, as someone not having a database of the web, I wouldn't be able to calculate either result.”
“It replied 29 resistors and provided a long list. Answer is 2 resistors”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.