DG.

This is Miko. She's a virtual streamer who is controlled by a real-life woman known only as The Technician. The Technician uses the Unreal Engine and a $30,000 motion-capture suit to create Miko. Thread 👇

We've evaluated a lot of base models on perplexity-based evals and Kimi k2.5 proved to be the strongest! After that, we do continued pre-training and high-compute RL (a 4x scale-up). The combination of the strong base, CPT and RL, and Fireworks' inference and RL samplers make Composer-2 frontier level. It was a miss to not mention the Kimi base in our blog from the start. We'll fix that for the next model.

Pakistan’s Domestic Solar Power Growth •2023 : 1 GW •2025 : 33 GW ⚡ Solar now contributes +25% of Pakistan’s total electricity

the value of this technology will mostly not be captured by its inventors, the labs, or even the chipmakers, but rather will be captured by the consumers as surplus. these are highly competitive markets without any natural monopolistic effects like many other technologies before it, machine intelligence democratizes abilities previously only available to the wealthy, in this case by commoditizing the services of the white collar elite who mostly live in rich countries it’s not that there are no programmers, it’s that really anybody can make software now now so the “rents” of the “human capital” of knowing how to write JavaScript for example should shrink dramatically this will reduce the inequality between countries: services that previously required lots of human capital now require chatbot subscriptions at worst, or may even be given away for free you can receive medical advice worthy of a $1000/hr American specialist doctor likely for free while living under a thatched roof in eg Papua New Guinea somewhere while I think Americans have plenty of reason to be excited by AI, I would be more excited as someone in a poor country

The U.S. has a weird cultural relationship with AI Despite the fact that we’ve driven the vast majority of AI breakthroughs, we still rank among the lowest countries in terms of consumer trust (Data from Edelman 2025 study) 👇

If you'd shown Claude Code or Codex on 5.4xhigh to any reasonable person in 2020 they'd have concluded it was AGI

Great thread. Speech to Speech models can mean a few things and all will behave differently. Turn based models won't interrupt you and duplex models will be constantly listening until the model decides to speak. Which model you use and the underlying behaviour (encoded at train time) determines what kind of applications are well suited for these models. You can't build a coaching app if your model doesn't know when to interrupt. It won't be natural. These are still new behaviours and tons of alpha left in exploring. These behaviours can't be prompted on the fly but need careful data curation. More startups should train models and explore horizontally. The best of e2e speech is still to come.