DeepSeek, a Chinese AI startup, recently published V3.2 and V3.2-Speciale, two reasoning models that are comparable to SOTA models like GPT-5 and Gemini 3 Pro while being affordable and available under an open-source license.

The specifics:

In math, tool usage, and coding assessments, V3.2 is on par with or close to GPT-5, 4.5 Sonnet, and Gemini 3 Pro, with the more demanding Speciale outperforming them in a number of areas.

The Speciale variation placed No. 10 overall at IOI and won gold in the 2025 International Math Olympiad and Informatics Olympiad.

In comparison to Gemini 3 Pro ($2 / $12), GPT-5.1 ($1.25 / $10), and Sonnet 4.5 ($3 / $15), V3.2 cost is $0.28 input / $0.42 output for 1 million tokens.

Both 685B parameter models are sold under an MIT license, and anyone can obtain the weights from Hugging Face.

DeepSeek's V3.2 follow-up demonstrates that the Chinese lab isn't a one-hit wonder by open-sourcing a model with frontier performance at a significant price decrease. DeepSeek's R1 release shook markets and raised discussions about U.S. chip export control. There is now much more pressure on U.S. labs who charge premium API rates to explain that disparity.

Your one-stop shop for automation insights and news on artificial intelligence is EngineAi.
Did you like this article? Check out more of our knowledgeable resources:
📰 In-depth analysis and up-to-date AI news .
🤝 Visit to learn about our goal and knowledgeable staff.
📬 Use this link to share your project or schedule a free consultation.
Watch this space for weekly updates on digital transformation, process automation, and machine learning. Let us assist you in bringing the future into your company right now.