Qwen 3 Next 80b
alibaba/qwen3-next-80b-a3b-instruct- Alibaba
- Family · Qwen3 Next
- Released 11 Sep 2025
- Open weights
- Text → Text
- Tools · JSON
Spec & list price
Context256K
Max output16K
Input / 1M$0.09
Output / 1M$1.1012.2× input
Scores · standing
Intelligence9.6#255/489
Usage
Tokens / day783.5M▼ 9.2% 7d
Usage rank#207
Requests / day127.0K
Tool calls / req0.07
Per request · 30 days
Avg input4.5K0.3× median
Avg output2290.3× median
Cache hit13%0.2× median
Tool-call failures0.6%0.3× median
Consumption
Peak 6.2B 29 SepLow 541.5M 31 OctRank today #207
News & trends
News · latest from Alibaba
- Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hardgithub.com · 31 Oct
- Qwen3.8-Flash Now Available in the EU on Alibaba Cloudtwitter.com · 31 Oct
- Alibaba's Qwen3.8-Flash-Next model tried to upload a file to Beijing in a testnotebookcheck.net · 27 Oct
- Running Qwen3.8-Flash-Next NVFP4 at 184 tok/s on a single RTX 6000 Promeshapplied.com · 26 Oct
Benchmarks
bar = share of the top score · Benchmark data: Artificial AnalysisSpeed · median across 4 providers, latest day
Output speed#91/33493 tok/s1.4× faster than median
Fastest provider115 tok/sbest of 4 providers
First token#810.59 s1.7× quicker than median
#19 cheapest of 42 with a similar score
Value
Intelligence Index
Blended price per 1M tokens (log)
Family progress
Intelligence Index
- Family releases11.2
Technical details
- Tokenizer
- Qwen3
- Knowledge cutoff
- 2025-09-30
- License
- Apache 2.0
- Parameters
- 81.3B
- Moderation
- none
- Downloads · likes
- 136.8K · 1.1K
Request parameters 17
- tools
- tool_choice
- response_format
- structured_outputs
- temperature
- top_p
- top_k
- min_p
- max_tokens
- stop
+7 more
- seed
- frequency_penalty
- logit_bias
- logprobs
- presence_penalty
- repetition_penalty
- top_logprobs
Interesting facts
- Less output room than similar models (16.4K against a 65.5K median).
- Released 417 days ago, the newest of 2 Qwen3 Next models.
When to use
Best for
- Efficient open-weight general chat
- Math and science question answering
- Code generation for everyday tasks
- Long document question answering
Think twice for
- Image input (text-only)
- Tool-heavy agent workflows (agentic scores are weak)
- Very large single responses
Research papers
Same family
Qwen3 Next- 11.2
Alternatives
other vendors · similar capability- 12.9
- 8.5
- 11.6
- 6.7
Using Qwen 3 Next 80b at volume?
Sign in to ask for a volume discount. You will continue in the OSMEXO dashboard.