1989
Amit Joshi says that when he asked DeepSeek about the 1989 Tiananmen Square events, it answered that nothing happened.
“it says, basically, I mean, I'm I'm paraphrasing here, but it said nothing happened. It was a nice, sunny day.”
Topic
DeepSeek
29 facts from 16 episodes, in date order. These are what speakers said on the shows, each linked to the moment it was said. They aren't independently verified.
2025
Tim Crawford said OpenAI had begun sharing some of its models' reasoning, and he thought DeepSeek was driving that change.
“We just saw, um, OpenAI come out and start to share some of the reasoning, and I think DeepSeek is driving some of this.”
2025
Dines says UiPath uses models from DeepSeek, Llama and Anthropic, among others, in different parts of its business.
“look at everything from deep seek, Lama, Anthropic, we use all of them in different parts of the business.”
2025
Dines says fine-tuning can distill an open-source model into a smaller model that works well for one domain, lowering cost.
“with fine tuning, you can actually distill it into a smaller model that works very well for a particular domain.”
2025
Jeetu Patel says Cisco jailbroke the DeepSeek model within 48 hours across 50 categories of the HarmBench benchmark.
“when DeepSeek came out, it only took us forty eight hours to jailbreak the model in fifty top categories in the harm bench benchmark, right?”
2025
Jeetu Patel reported a 100% attack success rate in Cisco's jailbreak testing of DeepSeek.
“And that attack success rate of 100% is very disturbing, um, because that's one time it's not good.”
2025
Daniel Whitenack says reasoning models such as OpenAI's o1 and DeepSeek R1 work by the same token-generation mechanism as ordinary language models.
“these quote reasoning models. In terms of the mechanism under which they operate, are exactly the same as what we just talked about. They produce tokens, probable tokens”
2025
Amit Joshi says DeepSeek is an open-source model whose weights can be modified and which can be run on a company's own servers.
“the leading Chinese model that's out there right now, which is DeepSeek, is an open-source model, which basically means I can take it, I can modify the weights, and I can put it on my servers”
2025
Amit Joshi says BMW plans to integrate DeepSeek into its vehicles for the Chinese and wider Asian markets.
“It's integrating DeepSeek, which is obviously the Chinese, uh, large language model, into its vehicles. So a lot of the AI in BMW vehicles is gonna be done by, uh, DeepSeek.”
2025
Wolf says DeepSeek's January release drew major adoption and visibly shook the stock markets.
“one that I think got the most adoption and most visibility recently was DeepSeek, who was released in January and kinda crashed the stock markets at the same time that it was released.”
2025
DeepSeek released its V3 model in December 2024 and R1 in January 2025, reaching frontier capabilities on a small budget according to Carvalho.
“the release of DeepSeek's LLMs, V3 in December 2024, R1, January 2025. These models developed frontier capabilities on a shoestring budget”
2025
Carvalho says DeepSeek reportedly trained its models without advanced NVIDIA GPUs, whose export was restricted under the Chips Act.
“supposedly without the advanced NVIDIA GPUs whose export was restricted under the Chips Act.”
2026
Flo Crivello says Lindy's default model is currently DeepSeek for everything, though the company is reconsidering because Teammate needs more compute.
“Everything is DeepSeek right now. We are in the middle of reconsidering it”
2026
Flo Crivello says DeepSeek Flash is roughly Sonnet 4.6 level and about a hundred times cheaper.
“Sonnet 4.6 is a really good level, a really good model for like most use cases. And it's literally a hundred x cheaper, you know?”
2026
Flo Crivello argues that Chinese frontier model makers distill American models, which he says creates unfair competition.
“Or number one, they're obviously distilling. It's very clear. And so you're putting American open source model companies in an unfair competition”
2026
Flo Crivello says that when he asks his own product about Tiananmen, the Chinese model refuses to answer, which he cites as evidence of censorship.
“When I talk to my own product and I'm asking it what happened in Tiananmen, it tells me, I'm sorry, I can't talk about that.”
2026
The host proposes an insurance requirement, with higher premiums for companies running less-understood Chinese models, as an alternative to banning them.
“And then I also wonder about kind of a market mechanism, like maybe insurance should, instead of a ban, like, what about an insurance requirement?”
2026
Ian Quigley says DeepSeek V4 Flash performs reasonably well on Leash Bio's benchmark.
“They're, uh, Deepseek V four Flash is pretty good, not that bad.”
Undated
Sridhar Ramaswamy says Snowflake hosted the DeepSeek model very quickly after its release.
“Things like hosting DeepSeek quickly, you know, that's just a little bit of making sure that you can still run the hundred meter sprint in ten seconds.”
Undated
Sridhar Ramaswamy says Snowflake made DeepSeek available to customers within two days.
“It was an amazing model. We had it out in two days flat.”
Undated
Sridhar Ramaswamy says DeepSeek's own app sends everything users type to servers in China.
“DeepSeek also offers services on servers that are hosted in China, where if you use their app, for example, everything that you are typing in is getting sent to China.”
Undated
Sridhar Ramaswamy says Snowflake's hosting of the DeepSeek model introduces no security compromise.
“Hosting the DeepSeek model does not introduce any kind of security compromise.”
Undated
Sualeh Asif says Cursor chose DeepSeek models as a pre-trained base for its own post-training and reinforcement learning stack.
“for just picking a really well pre trained base, uh, DeepSeek does a phenomenal, phenomenal job.”
Undated
Mollick said DeepSeek's R1 was the first Chinese AI model to reach the frontier of capabilities, though still a few months behind American models.
“DeepSeek R1 specifically, was it was the first time a Chinese model was on the frontier of capabilities.”
Undated
David Luan says DeepSeek surprised the world because it jumped directly to the reasoning-model training recipe without every intermediate step.
“That's why DeepSeek was able to surprise the world, because they straight quantum tunneled to that instead of doing every stop along the way.”
Undated
Daniel Whitenack names Llama, DeepSeek and Qwen as examples of open models, and GPT and Anthropic's models as examples of closed models.
“There's closed models, there's open models, like closed models being the GPTs or Anthropic or et cetera, the open models being the you know, Llama or Deepseek or Quinn or whatever.”
Undated
Snowflake offers DeepSeek among the models available to its customers.
“We also have DeepSeek as a model that we provide for customers.”
Undated
Daniel Whitenack uses DeepSeek as an example of a vendor that builds and trains a model.
“let's say a deepseek model. So let's say that you have a deepseek model. There is, uh, uh, an LLM or a vision model, or whatever it is, um, that this is a model that was built or trained by a particular vendor.”
Undated
AWS Bedrock can run a DeepSeek model on its own servers as a managed service.
“This would be like AWS Bedrock runs a DeepSeek model in their servers.”
Episodes
- 26 Oct 2026⚡ An Eval for AI Scientists That Knows the Answer Key - Ian Quigley, Leash Bio16m
- 16 Sep 2025Challenging the Average With Open-Source AI: Hugging Face’s Thomas Wolf30m
- 26 May 2025Are You Making These COMMON AI Organization Mistakes? | CXOTalk #88153m
- 18 Feb 2025Building trust in AI: A CIO's perspective | CXOTalk #86956m
- 12 Aug 2025Are You Considering Chinese AI In Your Strategy?32m
- 21 Aug 2025Amazon is betting on agents to win the AI race59m
- 7 Apr 2025UiPath CEO Daniel Dines on AI agents replacing our jobs1h 13m
- 11 Jun 2025Ethan Mollick: AI Urgency, Leadership Responsibility, and Why Companies Must Act Now39m
- 1 Apr 2025From data breach scandal to AI darling, with Snowflake’s CEO Sridhar Ramaswamy35m
- 19 Mar 2026295. AI, strategy, and the future of work: What business leaders need to know35m
- 29 Apr 2025Inside Cursor: The future of AI coding with Co-founder Sualeh Asif49m
- 14 Aug 2026Lindy Teammate: Flo Crivello on Multiplayer Agents, Memory & Why He'd Ban the Chinese Models He Uses2h 6m
- 14 Jan 2026Snowflake VP of AI Baris Gultekin on Bringing AI to Data, Agent Design, Text-2-SQL, RAG & More1h 36m
- 7 May 2026The Myth of Model Wars: Open vs Closed AI in 202642m
- 11 Sep 2025Cracking the code of failed AI pilots46m
- 7 Jul 2025AI in the shadows: From hallucinations to blackmail44m