Qwen3 30B A3B
alibaba/qwen3-30b-a3b- Alibaba
- Family · Qwen3
- Released 28 Apr 2025
- Open weights
- Text → Text
- Tools · JSON
Spec & list price
Context40K
Max output16K
Input / 1M$0.12
Output / 1M$0.504.2× input
Scores · standing
Intelligence7.6#324/490
Usage
Tokens / day441.2M▲ 47% 7d
Usage rank#233
Requests / day148.9K
Tool calls / req0.12
Per request · 30 days
Avg input2.3K0.2× median
Avg output5490.8× median
Reasoning share76%1.3× median
Tool-call failures0.4%0.2× median
Consumption
Peak 753.5M 26 SepLow 163.4M 10 OctRank today #233
News & trends
201 mentions this week ▲ 49%Qwen3 family · videos, posts, articles · 30 days
Trending · 14 days · Qwen3 family
- Qwen3.8 27b (Q6) casually breaking reCAPTCHA838 upvotes
- Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hard416 points
- Neuron Explorer for Qwen3.5-4B: See every neuron at work, make targeted edits26 points
- New Card for Qwen3.8 27b and MiniMax H3656 upvotes
- Qwen 3.8 27B vs Strata Coder - Local LLM battle75.2K views
- DavidAU Qwen3.8 27B Turbo Brilliance 50X-Reasoning... tested - Local LLM setup87.7K views
News · latest from Qwen3
- Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hardgithub.com · 31 Oct
- Neuron Explorer for Qwen3.5-4B: See every neuron at work, make targeted editsgithub.com · 26 Oct
- How to train your own Jev for $17Together AI · 23 Sep
Benchmarks
bar = share of the top score · Benchmark data: Artificial AnalysisSpeed · median across providers, latest day
Output speed#65/33498 tok/s1.6× faster than median
First token#810.61 s1.8× quicker than median
#12 cheapest of 40 with a similar score
Value
Intelligence Index
Blended price per 1M tokens (log)
Family progress
Intelligence Index
- Family releases33.7
Technical details
- Tokenizer
- Qwen3
- Knowledge cutoff
- 2025-03-31
- License
- Apache 2.0
- Parameters
- 30.5B
- Moderation
- none
- Downloads · likes
- 1.0M · 979
Request parameters 17
- tools
- tool_choice
- response_format
- structured_outputs
- reasoning
- include_reasoning
- temperature
- top_p
- top_k
- min_p
+7 more
- max_tokens
- stop
- seed
- frequency_penalty
- logit_bias
- presence_penalty
- repetition_penalty
When to use
Best for
- Cost-efficient reasoning on text tasks
- Self-hosted assistants with fast inference
- Bulk classification and extraction
- Math-oriented tutoring and problem solving
Think twice for
- Very long documents
- Image understanding
- Autonomous terminal-style coding agents
- Hardest scientific reasoning
Research papers
- End-to-End MXFP8 Reinforcement Learning on Blackwell: What Matching Buys, When It Pays Off, and How It Fails26 Oct
- SPARE: Sparsity Allocation via Routing Entropy for Efficient MoEs21 Oct
- Reconstruction-Optimized Expert Pruning for Mixture-of-Experts Language Models19 Oct
- Triple-Bottom-Line Sustainability of Language Models for Edge AI: A Comparison Between SLMs and Quantized LLMs1 Sep
- RotaryQuant: Fitting 120B MoE Models on Consumer Hardware via Fused Compressed-Space Attention8 Aug
- Shape Mutating Expert Compression:LorExperts and BTExperts7 Aug
- Open-SWE-Traces: Advancing Dual-Mode Multilingual Distillation for Software Engineering Agents14 Jun
- ACC: Compiling Agent Trajectories for Long-Context Training21 May
Same family
Qwen3- 7.2
- 6.0
- 6.9
- 6.1
- 5.2
- 13.1
Alternatives
other vendors · similar capability- 8.5
- 11.6
- 6.7
- 9.7
Using Qwen3 30B A3B at volume?
Sign in to ask for a volume discount. You will continue in the OSMEXO dashboard.