China's DeepSeek launches new AI models to rival ChatGPT, Gemini
What's the story
Chinese artificial intelligence (AI) start-up DeepSeek has launched two new models, the DeepSeek-V3.2 and DeepSeek-V3.2-Speciale. The move comes as a direct challenge to the dominance of Google's Gemini series and OpenAI's ChatGPT in the AI chatbot space. The company had earlier shot to fame with its DeepSeek R1 and V3 models that matched ChatGPT's performance at a fraction of the cost.
Model capabilities
DeepSeek-V3.2: A reasoning-first model
The new DeepSeek-V3.2 is a 'reasoning-first model built for agents,' according to the company. It claims that this model delivers performance on par with GPT-5 for general tasks. The other variant, DeepSeek-V3.2-Speciale, is said to have reasoning capabilities comparable to Google's latest Gemini 3 Pro model, especially in complex problem-solving scenarios.
Competition success
DeepSeek-V3.2-Speciale's performance in international competitions
The DeepSeek-V3.2-Speciale model has also shown exceptional performance in international competitions like the International Mathematical Olympiad (IMO) and International Olympiad in Informatics (IOI). It achieved Gold-medal level performance, a feat previously only achieved by proprietary models from Google and OpenAI. The model outperformed both GPT-5 High and Gemini 3 Pro on pure math benchmarks such as AIME and HMMT, and also excelled at pure coding logic tasks.
Benchmark performance
DeepSeek-V3.2-Speciale's performance on humanity's last exam
On the Humanity's Last Exam (HLE), one of the toughest benchmarks, DeepSeek-V3.2-Speciale outperformed GPT-5 High but fell short of Gemini 3 Pro. The new models are now available for use on the DeepSeek website and app. However, it's worth noting that DeepSeek V3.2-Speciale is only available for API use at this stage with no word on when it will be accessible to regular users.