Home
DeepSeek’s next model may feature 3 trillion parameters, targeting GPT-6 Astra performance

DeepSeek recently launched V4.1 Flash, introducing architectural shifts substantial enough to resemble a V5 release, yet the company opted to retain the existing version number. While the Pro series update, V4.1 Pro, has been confirmed, detailed specifications remain undisclosed. Assuming improvements mirror the V4 to V4.1 transition, V4.1 Pro could achieve exceptional performance; however, given the underwhelming reception of the previous V4 Pro-0813 final build, market expectations should remain measured.
DeepSeek Code 2.0 Imminent, Prioritizing Computer Control
Industry reports suggest DeepSeek Code 2.0, slated for a September release, will outperform current market leaders Mythos 5.1 and GPT-6 Astra. This next-generation model aims to surpass the capabilities of today’s most advanced large language models.
DeepSeek Code 2.0 is expected to maintain its open-source commitment, with a primary design focus on Computer Use. This capability enables AI to autonomously operate computers, executing complex workflows on behalf of users—a feature that has significantly enhanced the utility of models like Astra.
3 Trillion Parameters: China’s Largest Scale
The model reportedly features over 3 trillion parameters, a key driver for its enhanced performance. For context, China’s current strongest models include K3 with 2.8 trillion parameters, Qwen 3.8 Max with 2.4 trillion, and DeepSeek V4 Pro with 1.6 trillion.
Given that V4.1 Flash adopted a new architecture with doubled parameters, the leap to 3 trillion for DeepSeek Code 2.0 appears plausible. Competing with giants like Mythos 5.1 and Astra necessitates this scale; without it, achieving a higher performance ceiling would be challenging.
Nevertheless, the credibility of this leak remains uncertain. DeepSeek previously released three Code models between 2023 and 2024, which delivered modest results. If DeepSeek now positions a code- and logic-focused Code model as its flagship, it would strategically reserve V4.1 Pro and Flash for general agent tasks, marking a logical evolution in their product lineup.
Related article
U.S. Stocks Hit Historic Milestone as AI and Aerospace Giants Prepare for Trillion-Dollar Debut
Elon Musk, Sam Altman, and Dario Amodei, three titans of the technology sector, are advancing toward initial public offerings for their respective ventures. With SpaceX, OpenAI, and Anthropic—three industry behemoths nearing trillion-dollar valuation
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur
Google Tests Remy AI Agent for Gemini as Focus Shifts to User Control
According to Business Insider, Google is testing Remy, a new AI personal agent for Gemini. This tool aims to execute tasks on behalf of users, streamlining both professional workflows and daily routines.Currently, Remy is undergoing testing in an int
Related Special Topic Recommendations
Comments (0)
0/500

DeepSeek recently launched V4.1 Flash, introducing architectural shifts substantial enough to resemble a V5 release, yet the company opted to retain the existing version number. While the Pro series update, V4.1 Pro, has been confirmed, detailed specifications remain undisclosed. Assuming improvements mirror the V4 to V4.1 transition, V4.1 Pro could achieve exceptional performance; however, given the underwhelming reception of the previous V4 Pro-0813 final build, market expectations should remain measured.
DeepSeek Code 2.0 Imminent, Prioritizing Computer Control
Industry reports suggest DeepSeek Code 2.0, slated for a September release, will outperform current market leaders Mythos 5.1 and GPT-6 Astra. This next-generation model aims to surpass the capabilities of today’s most advanced large language models.
DeepSeek Code 2.0 is expected to maintain its open-source commitment, with a primary design focus on Computer Use. This capability enables AI to autonomously operate computers, executing complex workflows on behalf of users—a feature that has significantly enhanced the utility of models like Astra.
3 Trillion Parameters: China’s Largest Scale
The model reportedly features over 3 trillion parameters, a key driver for its enhanced performance. For context, China’s current strongest models include K3 with 2.8 trillion parameters, Qwen 3.8 Max with 2.4 trillion, and DeepSeek V4 Pro with 1.6 trillion.
Given that V4.1 Flash adopted a new architecture with doubled parameters, the leap to 3 trillion for DeepSeek Code 2.0 appears plausible. Competing with giants like Mythos 5.1 and Astra necessitates this scale; without it, achieving a higher performance ceiling would be challenging.
Nevertheless, the credibility of this leak remains uncertain. DeepSeek previously released three Code models between 2023 and 2024, which delivered modest results. If DeepSeek now positions a code- and logic-focused Code model as its flagship, it would strategically reserve V4.1 Pro and Flash for general agent tasks, marking a logical evolution in their product lineup.
U.S. Stocks Hit Historic Milestone as AI and Aerospace Giants Prepare for Trillion-Dollar Debut
Elon Musk, Sam Altman, and Dario Amodei, three titans of the technology sector, are advancing toward initial public offerings for their respective ventures. With SpaceX, OpenAI, and Anthropic—three industry behemoths nearing trillion-dollar valuation
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur











