Interconnects

Richard Bian, Product & Growth Lead at Ant Ling, shares insights on Ant Group’s vision for open language models. He discusses strategic openness, aiming to accelerate learning and avoid mistakes. Alongside algorithm engineer Chen Liang, they dissect the challenges of training stability and FP8 optimization. The conversation touches on the rapidly evolving Chinese AI ecosystem and the influence of DeepSeek on their mission. They explore balancing model sizes and the innovations behind their unique training methods, all while emphasizing a collaborative approach to AI.

First, congrats to the Moonshot AI team, one of the 6 “AI Tigers” in China, on the awesome release of Kimi K2 Thinking. One of the overlooked and inspiring things for me these days is just how many people are learning very quickly to train excellent AI models. The ability to train leading AI models and distribute them internationally is going to be pervasive globally. As people use AI more, those who can access supply for inference (and maybe the absolute frontier in scale of training, even if costly) is going to be the gating function.K2 Thinking sounds like a joy to use because of early reports that the distinctive style and writing quality from their original Kimi K2 Instruct model have been preserved through extended thinking RL training. They released many evaluation scores, for a highlight they’re beating leading closed models on some benchmarks such as Humanity’s Last Exam or BrowseComp. There are still plenty of evals where GPT 5 or Claude Sonnet 4.5 tops them. Rumors are Gemini 3 is coming soon (just like the constantly pending DeepSeek V4), so expectations are high on the industry right now.TLDR: Kimi K2 Thinking as a reasoning MoE model with 1T total, 32B active parameters, 256K context length, interleaved thinking in agentic tool-use, strong benchmark scores and vibe tests.The core reaction of this release is people saying this is the closest open models have been to the closed frontier of performance ever, similar to DeepSeek R1‘s fast follow to o1. This is pretty true, but we’re heading into murky territory because comparing models is harder. This is all advantaging the open models, to be clear. I’ve heard that Kimi’s servers are already totally overwhelmed, more on this soon.What is on my mind for this release:1. Open models release faster. There’s still a time lag from the best closed to open models in a few ways, but what’s available to users is much trickier and presents a big challenge to closed labs. Labs in China definitely release their models way faster. When the pace of progress is high, being able to get a model out sooner makes it look better. That’s a simple fact, but I’d guess Anthropic takes the longest to get models out (months sometimes) and OpenAI somewhere in the middle. This is a big advantage, especially in comms, to the fast mover.I’d put the gap at the order of months in raw performance — I’d say 4-6+ months if you put a gun to my head and made me choose specifically — but the problem is these models aren’t publicly available, so do they matter?2. Key benchmarks first, user behaviors later. Labs in China are closing in and very strong on key benchmarks. These models also can have very good taste (DeepSeek, Kimi), but there is a long-tail of internal benchmarks that labs have for common user behaviors that Chinese labs don’t have feedback cycles on. Chinese companies will start getting these, but intangible’s are important to user retention.Over the last year+ we’ve been seeing Qwen go through this transition. Their models were originally known for benchmaxing, but now they’re legitimately fantastic models (that happen to have insane benchmark scores).Along these lines, the K2 Thinking model was post-trained natively with a 4bit precision to make it far more ready for real serving tasks (they likely did this to make scaling RL more efficient in post-training on long sequences too):To overcome this challenge, we adopt Quantization-Aware Training (QAT) during the post-training phase, applying INT4 weight-only quantization to the MoE components. It allows K2 Thinking to support native INT4 inference with a roughly 2x generation speed improvement while achieving state-of-the-art performance. All benchmark results are reported under INT4 precision.It’s awesome that their benchmark comparisons are in the way it’ll be served. That’s the fair way.3. China’s rise. At the start of the year, most people loosely following AI probably knew of 0 Chinese labs. Now, and towards wrapping up 2025, I’d say all of DeepSeek, Qwen, and Kimi are becoming household names. They all have seasons of their best releases and different strengths. The important thing is this’ll be a growing list. A growing share of cutting edge mindshare is shifting to China. I expect some of the likes of Z.ai, Meituan, or Ant Ling to potentially join this list next year. For some of these labs releasing top tier benchmark models, they literally started their foundation model effort after DeepSeek R1. It took many Chinese companies only 6 months to catch up to the open frontier in ballpark of performance, now the question is if they can offer something in a niche of the frontier that has real demand for users.4. Interleaved thinking on many tool calls. One of the things people are talking about with this release is how Kimi K2 Thinking will use “hundreds of tool calls” when answering a query. From the blog post:Kimi K2 Thinking can execute up to 200 – 300 sequential tool calls without human interference, reasoning coherently across hundreds of steps to solve complex problems.This is one of the first open model to have this ability of many, many tool calls, but it is something that has become somewhat standard with the likes of o3, Grok 4, etc. This sort of behavior emerges naturally during RL training, particularly for information tanks, when the model needs to search to get the right answer. So this isn’t a huge deal technically, but it’s very fun to see it in an open model, and providers hosting it (where tool use has already been a headache with people hosting open weights) are going to work very hard to support it precisely. I hope there’s user demand to help the industry mature for serving open tool-use models.Interleaved thinking is slightly different, where the model uses thinking tokens in between tool use call. Claude is most known for this. MiniMax M2 was released on Nov. 3rd with this as well! It’s new.5. Pressure on closed American labs. It’s clear that the surge of open models should make the closed labs sweat. There’s serious pricing pressure and expectations that they need to manage. The differentiation and story they can tell about why their services are better needs to evolve rapidly away from only the scores on the sort of benchmarks we have now. In my post from early in the summer, Some Thoughts on What Comes Next, I hinted at this:This is a different path for the industry and will take a different form of messaging than we’re used to. More releases are going to look like Anthropic’s Claude 4, where the benchmark gains are minor and the real world gains are a big step. There are plenty of more implications for policy, evaluation, and transparency that come with this. It is going to take much more nuance to understand if the pace of progress is continuing, especially as critics of AI are going to seize the opportunity of evaluations flatlining to say that AI is no longer working.Are existing distribution channels, products, and serving capacity enough to hold the value steady of all the leading AI companies in the U.S.? Personally, I think they’re safe, but these Chinese models and companies are going to be taking bigger slices of the growing AI cake. This isn’t going to be anywhere near a majority in revenue, but it can be a majority in mindshare, especially with international markets.Interconnects is a reader-supported publication. Consider becoming a subscriber.This sets us up for a very interesting 2026. I’m hoping to make time to thoroughly vibe test Kimi K2 Thinking soon!Quick links:* Interconnects: Kimi K2 and when “DeepSeek Moments” become normal, China Model Builder Tier List (they’re going up soon probably)* Model: https://huggingface.co/moonshotai/Kimi-K2-Thinking* API: https://platform.moonshot.ai/ (being hammered)* License (Modified MIT): The same as MIT, very permissive, but if you use Kimi K2 (or derivatives) in a commercial product/service that has >100M monthly active users or >$20M/month revenue, you must prominently display “Kimi K2” on the UI. Is reasonable, but not “truly open source.” https://huggingface.co/moonshotai/Kimi-K2-Thinking/blob/main/LICENSE* Technical blog: https://moonshotai.github.io/Kimi-K2/thinking.html* Announcement thread: https://x.com/Kimi_Moonshot/status/1986449512538513505 This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe

The conversation kicks off by dissecting the intense work culture in AI, contrasting performative overwork with the grim realities it spawns. Comparisons to historical industry pressures reveal high stakes, including mental strain and even fatalities. As researchers clock 100-hour weeks, the discussion shifts to the balance of rest and creativity. Nathan highlights talent constraints as the new bottleneck in progress. With the ongoing race for marginal gains, he warns that only the most committed will keep pace, emphasizing the need for long-term vision and self-care.

Explore the exciting world of scaling reinforcement learning as Nathan dives into the challenges and opportunities ahead. Discover the groundbreaking ScaleRL paper, which predicts learning curves and outlines the critical constants influencing RL performance. Learn how recent algorithmic advancements, like truncated importance sampling, are revolutionizing the field. Plus, gain insights into Pipeline RL's systems improvements that minimize GPU idle time. This is a journey into refining RL experimentation and boosting efficiency!

Explore the exciting shifts in the open model landscape from 2024 to 2025! Discover how China's models gained cumulative lead and the rise of Qwen, alongside Llama's strategic retreat. Delve into the implications of fine-tuned variants and community engagement. Concerns over security and the call for a transparent U.S. response add to the urgency. Plus, audience Q&As reveal insights on the competition between U.S. and Chinese models and the future of open-source initiatives. It's a fascinating look at where AI is headed!

A recent conference sparked vibrant debates on AI timelines and progress. Key discussions included the plausible automation of research engineers within 3–7 years. Experts are concerned about assuming a certain sequence of AI development is inevitable. The complexity of models and tools was explored, revealing both potential productivity boosts and challenges. Predictions suggest significant advancements and stagnations in AI capabilities ahead, while geopolitical implications of model standards were highlighted. The urgent need for proactive policies regarding open models was also emphasized.

The podcast dives into the monetization of ChatGPT, exploring the challenges of shopping and advertising. Special focus is on the recent launch of the 'Buy It' feature and the innovative Agentic Commerce Protocol. There’s a discussion about the precision needed for AI to shop effectively and the advances in search capabilities that make this possible. It also highlights the growing trend of model specialization and the potential of agentic apps to transform daily tasks. Caution is advised regarding AI's impact on marketplaces, yet optimism shines through.

Explore the evolution of reasoning models and how they've outgrown the limitations of early AI like ChatGPT. Delve into the three core primitives: thinking, searching, and acting, each pivotal for future advancements. Uncover the nuances of hallucinations framed as context problems and the ongoing debate between open and closed tool ecosystems. Nathan highlights the impact of tokenomics and hardware trends on AI efficiency, emphasizing that these innovations are set to revolutionize our interaction with technology.

Coding is identified as the last tractable frontier for AI, providing consistent improvements that are highly valuable. Unlike chat and mathematics, coding is proving to be the most useful domain for AI advancements. Real-world applications show agents as powerful tools that enhance workflows, acting as persistent editorial assistants. The discussion includes hands-on performance comparisons between coding agents and traditional methods, highlighting how they solve complex coding issues with ease, transforming the way we interact with technology.

Dive into China's ambitious open-source AI strategies and their quest for global influence. The podcast highlights government efforts to bolster national AI capabilities and the geopolitical ramifications of these developments. It discusses the pride within Chinese tech sectors, the ongoing pursuit of self-sufficiency, and the innovative spirit driving this evolution. Listeners can ponder the potential impacts of China's AI efforts on the international landscape thanks to initiatives like the AI Plus plan.