Deepseek V3.2 Exp: CHEAPEST Model Ever That Is Quite POWERFUL! Best Opensource Model?
This week in AI saw potential Gemini 3.0 leaks, GLM 4.6 release, Claude 4.5 Sonnet unveiling, and Deep Seek V3.2's cost-effective sparse attention model launch.
MAIN POINTS FROM TRANSCRIPT
- Gemini 3.0 leaks suggest a launch next week with early access available.
- GLM 4.6 and Claude 4.5 Sonnet introduce powerful new open-source and coding models.
- Deep Seek V3.2 reduces API costs by over 50% using sparse attention.
- Deep Seek V3.2 maintains performance while being significantly cheaper than proprietary models.
TAKEAWAYS
- Sparse attention in Deep Seek V3.2 enhances efficiency by focusing on important words.
- The experimental model tests improvements for future iterations like Deep Seek R2.
- Deep Seek V3.2 offers competitive performance at a fraction of the cost.
- Access to Deep Seek V3.2 is free via chatbot or API, with local hosting options available.