Kaching999
Galvanizer
mayank11280
Forerunner
I don't know about others but I am finding DeepSeek R1 (through Perplexity) responses better than GPT 4o for my use case. It is succinct and poignant.
mayank11280
Forerunner
No, no. On Perplexity you can use R1 reasoning model hosted in the US on Perplexity servers.
ChatGPT O3mini has been launched for free. O3mini (medium reasoning) beats the DeepSeek R1 in most benchmarks.
DeepSeek made OpenAI to launch its model for free. O3mini is even better than O1-pro, which costs $200/month.
DeepSeek made OpenAI to launch its model for free. O3mini is even better than O1-pro, which costs $200/month.
Thanks to techenclave, I came to know about Deepseek and installed android app and it's too good to interact, never felt same thing with google assistant and I like Deepseek using emojis.
Will be installing it locally, can you/anybody create a guide here.
Attachments
psyph3r
Juggernaut
Opinion: Opinion | DeepSeek Shows Why Nandan Nilekani Is Right
India's track record suggests that building AI models could run up against scaling and capital challenges. A better strategy might be what Infosys co-founder Nandan Nilekani proposed last year.
www.ndtv.com
Look no further than lmstudio.ai and ollama.com depending on how you want to run it locally... they should have their own docs, wiki, videos, user communities for your questions etc.
kartikoli
Innovator
I use jan.ai on desktop, its simple as installing any software on PC. Once installed select the models to use
Skoomahed
Contributor
One combination is : Open-webui + Ollama - This will give you same UI as chatgpt and comes with many useful features like uploading documents, web search, user access control etc
If you want to use LLMs for more creative stuff, like story writing, roleplaying etc then i recommend KoboldCPP + Sillytavern
Finance Ministry Bans ChatGPT, DeepSeek Over Data Risk
The Finance Ministry has banned its officers from using AI tools like ChatGPT and DeepSeek due to data confidentiality concerns. This follows similar bans in Australia and Italy. Read more.
For those that want an EASY way to run models locally on CPU / CUDA / ROCm, the path of least resistance I've found is KoboldCPP. Single executable, no dependencies, no setup.. just download whatever model you like from Huggingface that will fit on your hardware and go. Super simple, but still has tons of features for people like me that have been doing this since the ancient days of AI (about 2 years ago haha). Hit me up if you need any help. Might be slow to respond as got a lot of stuff going on, but happy to help when I get to it.
Ofc they likely have zero understanding that DeepSeek is a fully open model can be hosted locally, and even recreated locally, so there is no external data risk.. but yeah same knee-jerk stuff is happening the world over.Finance Ministry Bans ChatGPT, DeepSeek Over Data Risk
The Finance Ministry has banned its officers from using AI tools like ChatGPT and DeepSeek due to data confidentiality concerns. This follows similar bans in Australia and Italy. Read more.money.rediff.com
bssunilreddy
Keymaster
Paris AI summit next week will be the test for US, China cooperation
The Paris AI Summit will test whether the US and China can find common ground on AI governance or if their growing rivalry will escalate further.
The Paris AI Summit, taking place February 10-11, will bring together nearly 100 nations to discuss AI governance, national sovereignty, and open-source AI. The real test will be whether the US and China can work together.
As world leaders gather in France next week for the AI Action Summit over February 10-11, all eyes will be on President Trump to see whether the United States and China can find some common ground on the matter of AI governance. With nearly 100 nations in attendance, the event focuses on the safe development of artificial intelligence, with focus areas covering everything from labor disruptions to national sovereignty, global AI governance, and open-source AI
However, the real test will be whether the US and China can bridge the growing gap created by the AI arms race. With tensions rising in Washington over AI leadership and national security, stakeholders from tech companies including Microsoft, Google, and OpenAI are also slated to attend. It remains unclear whether DeepSeek CEO Lian Wenfeng will be present, though the company's rapid rise is certain to be a major talking point.
Beyond geopolitical tensions, funding for AI development is expected to be a likely outcome of the summit, with philanthropies and businesses expected to commit $2.5 billion over five years to public-interest AI projects. Creating new AI regulation is also not on the agenda; rather, the focus will be on refining existing frameworks. Clean energy will also be a focus, given the inevitable power consumption associated with powering AI models and the push for sustainable solutions.
Source: https://www.tweaktown.com/news/1030...-the-test-for-us-china-cooperation/index.html
AMD downplays DeepSeek threat: '[Innovation is] good for AI adoption'
In a recent Q4 earnings call, AMD's CEO Lisa Su provided some comments on hardware sales, demand for AI, and DeepSeek's effect on the broader industry.
Despite a $1 trillion tech sector crash, AMD CEO Lisa Su and other industry leaders praise DeepSeek’s AI efficiency, seeing it as a driver for broader AI adoption rather than a threat to hardware demand.
While markets respond to the shift in demand created by Chinese AI company DeepSeek, AMD's CEO Lisa Su provided some comments on the state of hardware sales moving forward.
"Relative to DeepSeek, we think that innovation on the models and the algorithms is good for AI adoption," she told analysts during Tuesday's Q4 earnings call.
Su welcomes the innovation, going on to state that she expects to see more AI compute deployment within the broader market.
"The fact that there are new ways to bring about training and inference capabilities with less infrastructure is actually good,"
Fears are still lingering within the tech sector, with power, semiconductor and infrastructure companies exposed to a collective $1 trillion loss in market valuation as of last Tuesday. NVIDIA suffered the most from this crash, famously losing $600 billion in market valuation and 17% in stock price over a single day.
AMD, however, has been somewhat resistant to the crash due to its relatively minor involvement in the AI sector. Since the DeepSeek R1 model launched on January 20th, AMD's stock price has decreased by just 2.3%, while NVIDIA is still down by approximately 15%.
Su also anticipates that AMD's next-gen Instinct accelerators will drive billions of dollars in annual revenue over the coming years, despite the speculation that less computing power will be required to produce next-gen AI models. Su was resistant to committing to a specific revenue figure for the accelerators.
AMD's CEO is not the only major tech company to praise DeepSeek recently, with the chiefs of Apple, Meta, Microsoft, and Google each issuing statements in favor of the innovation. The common sentiment is that lowering the cost of artificial intelligence will be a net positive for the industry, with AI adoption only expected to increase.
"First of all, I think, tremendous team," Google's Sundar Pichai said of DeepSeek. "I think they've done very, very good work."
Even as AI hardware giants suffer historic losses, top executives from AMD to Google are praising DeepSeek's efficiency as a positive shift for the industry. Whether this optimism will translate into long-term gains remains the burning question.
Source: https://www.tweaktown.com/news/1030...innovation-is-good-for-ai-adoption/index.html
The Paris AI Summit will test whether the US and China can find common ground on AI governance or if their growing rivalry will escalate further.
The Paris AI Summit, taking place February 10-11, will bring together nearly 100 nations to discuss AI governance, national sovereignty, and open-source AI. The real test will be whether the US and China can work together.
As world leaders gather in France next week for the AI Action Summit over February 10-11, all eyes will be on President Trump to see whether the United States and China can find some common ground on the matter of AI governance. With nearly 100 nations in attendance, the event focuses on the safe development of artificial intelligence, with focus areas covering everything from labor disruptions to national sovereignty, global AI governance, and open-source AI
However, the real test will be whether the US and China can bridge the growing gap created by the AI arms race. With tensions rising in Washington over AI leadership and national security, stakeholders from tech companies including Microsoft, Google, and OpenAI are also slated to attend. It remains unclear whether DeepSeek CEO Lian Wenfeng will be present, though the company's rapid rise is certain to be a major talking point.
Beyond geopolitical tensions, funding for AI development is expected to be a likely outcome of the summit, with philanthropies and businesses expected to commit $2.5 billion over five years to public-interest AI projects. Creating new AI regulation is also not on the agenda; rather, the focus will be on refining existing frameworks. Clean energy will also be a focus, given the inevitable power consumption associated with powering AI models and the push for sustainable solutions.
Source: https://www.tweaktown.com/news/1030...-the-test-for-us-china-cooperation/index.html
AMD downplays DeepSeek threat: '[Innovation is] good for AI adoption'
In a recent Q4 earnings call, AMD's CEO Lisa Su provided some comments on hardware sales, demand for AI, and DeepSeek's effect on the broader industry.
Despite a $1 trillion tech sector crash, AMD CEO Lisa Su and other industry leaders praise DeepSeek’s AI efficiency, seeing it as a driver for broader AI adoption rather than a threat to hardware demand.
While markets respond to the shift in demand created by Chinese AI company DeepSeek, AMD's CEO Lisa Su provided some comments on the state of hardware sales moving forward.
"Relative to DeepSeek, we think that innovation on the models and the algorithms is good for AI adoption," she told analysts during Tuesday's Q4 earnings call.
Su welcomes the innovation, going on to state that she expects to see more AI compute deployment within the broader market.
"The fact that there are new ways to bring about training and inference capabilities with less infrastructure is actually good,"
Fears are still lingering within the tech sector, with power, semiconductor and infrastructure companies exposed to a collective $1 trillion loss in market valuation as of last Tuesday. NVIDIA suffered the most from this crash, famously losing $600 billion in market valuation and 17% in stock price over a single day.
AMD, however, has been somewhat resistant to the crash due to its relatively minor involvement in the AI sector. Since the DeepSeek R1 model launched on January 20th, AMD's stock price has decreased by just 2.3%, while NVIDIA is still down by approximately 15%.
Su also anticipates that AMD's next-gen Instinct accelerators will drive billions of dollars in annual revenue over the coming years, despite the speculation that less computing power will be required to produce next-gen AI models. Su was resistant to committing to a specific revenue figure for the accelerators.
AMD's CEO is not the only major tech company to praise DeepSeek recently, with the chiefs of Apple, Meta, Microsoft, and Google each issuing statements in favor of the innovation. The common sentiment is that lowering the cost of artificial intelligence will be a net positive for the industry, with AI adoption only expected to increase.
"First of all, I think, tremendous team," Google's Sundar Pichai said of DeepSeek. "I think they've done very, very good work."
Even as AI hardware giants suffer historic losses, top executives from AMD to Google are praising DeepSeek's efficiency as a positive shift for the industry. Whether this optimism will translate into long-term gains remains the burning question.
Source: https://www.tweaktown.com/news/1030...innovation-is-good-for-ai-adoption/index.html
Understanding is that they are banning direct usage of online Chinese deepseek website for analyzing Indian finance data.
Using deepseek in local servers air-gapped from internet don't have to be an issue, as govt itself is testing the same.
Govt.'s concern can be about leakage of Indian financial data to Chinese which is a very valid point, though Dow Jones and other statistical monitoring organizations are doing the same deeply for decades.