Scaling the Next Paradigm of Heterogeneous Intelligence

Talk transcript

1. 0:00

   [upbeat music] Thank you for coming to my talk.

2. 0:16

   My name is Adrian Bertagnoli. I'm a founding engineer at Callosum, and today I'm gonna be talking about scaling the next paradigm of heterogeneous intelligence. So I'm gonna start, um, with explaining why we care about heterogeneity in the first place, what particular aspect make it very conducive for scaling AI, um, how it is actually used in practice today,

3. 0:40

   and how we can use-- utilize it in the future, um, to actually scale the next paradigm of intelligence. So to give you an intuition about what I mean with heterogeneous intelligence, I wanna take a step back and, um, explain the current prevailing paradigm of homogeneous, uh, intelligence.

4. 1:00

   So homogeneous intelligence in, in terms of AI mainly refers to scaling single models on a fleet of identical chips. So this was largely-- This era was largely brought about by the discovery of neural scaling laws, which showed us that more data and more parameters leads to better models.

5. 1:19

   However, this is primarily rooted in a training domain, and while we move towards an inference domain, this becomes less and less relevant. So it's already changing, um, and we already see some level of heterogeneity in-- um, going into our current systems.

6. 1:39

   So on the architecture level, we see that mixture of experts are replacing large dense models. On the workflow layer, we see that single LLM calls are being replaced by mi-- uh, multi-agent systems.

7. 1:53

   And finally, on the hardware level, single chips are being replaced by prefill decode disaggregated systems. So given that we are currently at the state of mild heterogeneity, how can you imagine, um, a, a greater level of heterogeneity?

8. 2:11

   How will that appear? So initially, will be what we currently are experiencing is mild heterogeneity. So everything is still running primarily on, on homogeneous clusters, but we have some variety in the prompts.

9. 2:27

   When we run, uh, multi-agent systems, we might use different LLMs for different sub-agents. Um, again, we have mixture of experts.

10. 2:36

    Uh, when we increase the heterogeneity, we might start to use different chips, um, for different models. So different LLMs might be put on different GPUs. Uh, they might be interacting.

11. 2:50

    We might be using different models completely. So we have a increase of state-space models, diffusion models, all interacting with each other, all on optimal hardware that exists currently. And the last stage, uh, where we really see the, the heterogeneous paradigm unfolding is when we have a co-evolution of systems, uh, hardware and software.

12. 3:14

    So th-there will be a unification where you'll have, um, a complete vertical integration of intelligence and hardware.

13. 3:24

    So why heterogeneity? Why is it a good thing in the first place? So real-world problems are complex, multi-step, and open-ended. They decompose into sub problems which require vastly different types of intelligences.

14. 3:40

    So scaling a singular type of intelligence to solve these is very inefficient and, and not optimal. So how do we solve them? Solving these actually requires models of different architectures and sizes working together, um, uh, acting together in long horizons, something we like to call multi-agent heterogeneous intelligence.

15. 4:05

    Furthermore, new generations of silicon is coming towards the market, but currently there's no interface which allows it to, um, this new hardware to be unified and, and constructively, um, help the current compute stack.

16. 4:20

    And so this is what we aim to change. So heterogene-heterogeneity, the benefit is not simply a belief that we have. We actually formalize it and proved it mathematically. Um, on the right-- on the left, you see a figure outlining, uh, the principle of maximum heterogeneity.

17. 4:42

    So these are heterogeneous agents where the color indicates, um, a distribution over a skill space. Um, if you take-- If you have a communication between these, here indicated by a ring topology, you can have a s-- what we like to call a production function.

18. 5:01

    And the production function is simply the demand, uh, can, can be well suited for the demand of one problem, but ill-suited for another problem. So here we have a, a production function that's well-suited for demand A and ill-suited for demand B.

19. 5:18

    If you want to do this in a homogeneous fashion, you would either be able to only scale one peak, or in the optimal case, to match this demand function, you would have only generalists, so as broad as possible, the skill set.

20. 5:31

    But then ultimately you'd have a very short cylinder that does not meet the production function readily. So we formalize this, and we, um, saw that across many domains, including neuroscience, economics, and ecology, these, these trends hold, and under any reasonable amount of constraints, heterogeneous systems outperform homogeneous ones.

21. 5:56

    So how do we use this in practice? Like, I've been telling you about the benefits of heterogeneity, but I've not told you anything about what it actually means in terms of AI.

22. 6:07

    So we optimize, um, multi-agent systems at three different parts of the workflow. So all the way from the hardware where agents run on, we choose different hardware depending on the computational demands on the agents, and then how agents interact and what workflow they construct.

23. 6:27

    Um, so we have already demonstrated multiple, uh, benefits of, of, uh, this type of orchestration, and I wanna go into a couple ones. Um, main- namely in the workflow, something, a primitive we like to call heterogeneous recursion, and in the agent layer, I wanna talk about multi-model, multi-modal video action language models.

24. 6:52

    So heterogeneous recursion. This is something, um... Who's here heard of recursive language models?

25. 7:01

    Okay. So for those of you who, um, know- don't know recursive language model, it's kind of a seminal paper that came out of MIT, uh, last October, and they basically showed that even if you only occupy a small, uh, percentage of the context window, you still can have dramatic context rot depending on the information complexity you want,

26. 7:23

    uh, from the prompt. So if you're doing a needle in a haystack task, that is O of one. The, the information requirement scales is constant throughout, regardless of how big the prompt is.

27. 7:36

    And then you can imagine adding up the rows. You, you, you give, uh, rows and columns. Adding up the rows would be O of N because as the prompt increases, the infor- informational requirement increases linearly.

28. 7:49

    So if the, if you have a constant information requirement, it scales well. You, you can occupy the full context window and actually get a, a good answer. However, when you go to linear or quadratic, uh, it degrades at around 60 to 30%.

29. 8:04

    So recursive language models solve this problem by actually treating the context as a, um, environment rather than putting it all into the prompt. So in practice, this looks like you present the context in a file, and then the, a coding agent interacts with it programmatically through Python REPL, um, basically doing keyword searches, regex, a- and other

30. 8:29

    tricks to extract subcontexts, and this subcontext is then passed off to an identical recursive agent. So this agent then can answer the question or spawn another recursive agent. And, and that's why it's called recursive, uh, language model.

31. 8:44

    So we simply extended this concept. Um, instead of using a single model on a single chip, we map based on the subcontext generated towards different chips and different models to emulate the performance while drastically, um, being cheaper and faster.

32. 9:06

    So here are results. Um, you can see this is on the OOLONG benchmark. This is basically, um, the benchmark they used in the paper. Um, and GPT-5, GPT-5.2 was the, um, most recent one when we produced this work.

33. 9:21

    Uh, it sits around here where it takes around two thousand seconds, um, to run through the, the benchmark, and it costs around, uh, $3.75 for one task. Our system, when we go on Cerebras, we s- are seven times cheaper and five times faster.

34. 9:41

    So you save incredibly much time or a lot cheaper, so it's basically like having your cake and eating it too. Um, with SamaNova, we even get, uh, further. We push the price down even further at the cost of some latency.

35. 9:57

    So we're 12 times cheaper and three times faster. So these are like making architectural decisions that are not like simply based on the hardware. You can make huge impactful, um, price differences and, and, and while m- emulating the intelligence you would have from frontier models.

36. 10:20

    So the next problem we wanted to address is basically, um, visual web navigation. So we used a mixture of open and closed, uh, video action language models, um, and we managed to beat, uh, the state-of-the-art of Video Web Arena, beating GPT-5.2 and Gemini, uh, 2.5 by 18% and 25% respectively.

37. 10:45

    And not only this, the way we did it is instead of treating the problem as a homogenous one, we, we acknowledge that the problem is heterogeneous, uh, i- itself.

38. 10:59

    It, it decomposes into multiple steps of visual reasoning, of, of textual reasoning, and each of these subcomponents requires different models to be, um, completed successfully. So here you see a fundamental shift of the Pareto frontier where you see singular models like Kimi K, uh, 2.5 and GPT-5.2

39. 11:23

    are outperformed by a mixture, a heterogeneous set of, of models. So when we use Qwen3-VL-8B-Instruct and Kimi K2.5, we're 1.3 times faster than using Kimi, Kimi alone.

40. 11:39

    We're 18 times cheaper than using, uh, GPT-5.2 alone. Um, and if we use, uh, Qwen3, um, plus GPT, we're actually three times faster and 3.7 times cheaper. So this is only benefit.

41. 11:54

    There's no downside, uh, in, in constructing this in a heterogeneous manner. So one part of, of our differentiating factor, why, how we were able to beat the state-of-the-art is- That we mapped certain subtasks like zooming and, and creating a different visual, um, reasoning for the agent.

42. 12:14

    We offloaded that into less intelligent models because you don't need GPT to zoom for you. So alone on these subtasks, we're able to be 11 times faster and 43 times cheaper than using ChatGPT.

43. 12:29

    And so this is what overal- overall accumulates towards these 3.7 times cheaper and three times faster.

44. 12:38

    So looking ahead, how do we s- view the future of compute? The first era of scaling compute was dominated by the CPU, where compute got quicker. The second era was making compute massively parallel.

45. 12:54

    This is dominated by NVIDIA. And the third paradigm,

46. 12:59

    compute is gonna become heterogeneous, mapping onto multi-agentic workloads and optimally, um, mapping these workloads onto different chips. We are actually, um, working with ARIA, the UK institute.

47. 13:16

    Um, we got a 3 million grant for the first-- for operating the first heterogeneous co-located cluster in the UK. So we really wanna make a difference and, and spearhead this new era of innovation.

48. 13:31

    So the era of h- homogeneous scale delivered extraordinary progress. We should be grateful for it. What comes next is heterogeneous intelligence, where models, workflows, and silicon co-evolve, and every new source of diversity makes the whole system smarter, faster and cheaper.

49. 13:48

    This is the worst our infrastructure will ever be. Thank you. [audience applauding]

50. 14:01

    How do you define which task to run on the, like, faster, uh, cheaper model? Like for instance, a zoom, is that something that's hard coded like, "Oh, if you have to zoom in this model," or do you have like a smarter model that decides the

51. 14:15

    So it is, um... Initially, we started doing bespoke decisions on mapping, uh, certain simple subtasks to simple models. But since then, we have, um, created an automation layer that detects the task complexity and automatically predicts the best model-- the best-suited model and hardware.

52. 14:41

    Any other questions? Great. Thank you so much for your attention. My name is Adrian Bertagnoli, and if anyone is interested, uh, we are hiring, so

53. 14:55

    yeah. Great. Thank you very much. [upbeat music]