A
Qwen3 30B A3B Thinking 2507OPEN
Qwen3-30B-A3B-Thinking-2507 is a 30B parameter Mixture-of-Experts reasoning model optimized for complex tasks requiring extended multi-step thinking.
Prices update automatically — checked daily against provider list prices.
- $0.75 per 1M tokens, assuming 3 input for 1 output the convention used by Artificial Analysis
- cheaper than 24% of the 191 open-weight models tracked here
- 38 of 51 Alibaba models tracked here, from cheapest
Benchmarks & performance
Independent benchmark scores for Qwen3 30B A3B Thinking 2507, measured by Artificial Analysis. Higher is better. Measurement mode: Reasoning.
For the same budget, Qwen3 30B A3B Thinking 2507 delivers 3.1 times more capability. Open weights also mean you can self-host it and pay nothing per token.
Models at this level cost $0.45 per 1M tokens (median of 18) — this one costs $0.75.
Cheaper and better on this index: GLM 5.3 Flash · Gemini 3.7 Flash (batch) · GPT-5.6 Luna and 65 more
About this model
Qwen3 30B A3B Thinking 2507 is an open-weight AI model by Alibaba. You can download and self-host it for free; the prices below are hosted-API list prices, tracked daily, for when you prefer convenience over self-hosting.
Frequently asked questions
What is Qwen3 30B A3B Thinking 2507?
Qwen3 30B A3B Thinking 2507 is an AI language model from Alibaba. It is open-weight: you can download it and run it on your own hardware, for free. It scores 14.6 on the Artificial Analysis intelligence index.
Is Qwen3 30B A3B Thinking 2507 free?
The weights are free and open — you can self-host Qwen3 30B A3B Thinking 2507 and pay nothing per token. If you prefer a hosted API, list prices are $0.2 per million input tokens and $2.4 per million output tokens.
What is Qwen3 30B A3B Thinking 2507 good at?
Independent benchmarks from Artificial Analysis give it GPQA 70.7%, MMLU-Pro 80.5%, Humanity's Last Exam 10.3%, Long Context Reasoning 59.7%, LiveCodeBench 70.7%, SciCode 33.3%, MATH-500 97.6%, AIME 90.7%, AIME 2025 56.3%, IFBench 50.7%, τ²-Bench 28.1%, τ-Bench Banking 5.4%, Terminal-Bench 1.5%, Terminal-Bench Hard 5.3%. It is particularly used for code generation.
Can I self-host Qwen3 30B A3B Thinking 2507?
Yes. Qwen3 30B A3B Thinking 2507 has open weights, so you can download it and run it on your own GPU or server with tools like Ollama, vLLM or llama.cpp — with no per-token cost.