JUST IN: Anthropic offically releases Mythos-class model Claude Fable to the public.

Mira Murati says human-AI collaboration needs models that can listen while they think: "The types of models that we work with today, they're very turn-based. You talk, they talk, then they go off and think." "While they're thinking, it's almost like they're deaf and blind. They cannot perceive anything else about what's going on." "By contrast, our interactions with each other are very rich. There is a lot of information in our interactions when we are silent, when we're thinking, when we're interrupting one another." "Interaction models are able to capture all of this nuance. They're not turn-based. They're more like time-based interaction, where they're continuously taking in audio, text, video, and continuously providing output." "This enables you to catch things like interruptions and simultaneous speech, and really create a rich, high bandwidth interaction between humans and machines." @miramurati at Bloomberg Tech live with @emilychangtv

reforge q1 2026 overview >5,054 companies/projects across private markets (fintech, ai, robotics, frontier tech) sourced through Ava >3,051 proceeded to deeper diligence and Ava's research stack produced an in-depth memo >202 teams received an email or DM from us; that's a 6% conversion from the prior diligence phase >20 of those were discussed at two or more ICs >2 investments made tldr: Top of funnel is healthy by quantity - more teams than ever building but quality is just not there; there's a lot of noise accompanying very little signal. Makes sense given the vibe coding craze and software commoditizing. So just because we decided not to participate in superfluous twitter discourse around fundraising in crypto or have our name put on a spreadsheet does not mean we are not extremely active. To us, the investable subset of opportunities is just extremely small. note 1: This does not include our inbound stats; that said, we execute a outbound-heavy strategy note 2: 3,051 -> 202 includes filtering out projects that may be too out-of-scope for us (robotics, frontier tech, etc.) but if there are signals that indicate a co could be generational, we'll still include in our pipeline reminder: Reforge is not just a "crypto vc". We are a venture capital and technology firm investing across the financial system value chain with a very broad mandate Some insights: Consensus is rising around core primitives but very few teams are actually solving the hard problems. the most crowded sectors are what you'd imagine and we’ve found difficult to build conviction in their long term sustainability: - x402/agentic payments/agent commerce rails - prediction markets: terminals, aggregators, wrappers - stablecoins/payments/neobanks - ai trading agents/execution wrappers Where we felt differentiation specifically in finance/crypto: - market-structure, risk-transfer, liquidity, and margin primitives - credit, settlement, and post-trade infrastructure - software-native markets with novel agent-workflow infrastructure (no wrappers!) With Ava, we have more data than ever on early-stage private markets across multiple sectors, predominately ai and finance. If you have any questions or are curious to learn more about what we’re seeing, just let us know

people suggesting what the hyperliquid core team should do with tokens is kind of funny, just let them cook if you were them you’d have already messed it up by now

SpaceX millionaires 4,000 x $1mil , 400 x $100 mil Every employee who joined before the first succesful launch made (unless they sold early) more than $100 million. SpaceX lists June 12 at ~$1.75T. Work backward from the cap table. At $1.75T, clearing $100M takes ~0.0057% of the company. - 2002–2008, first ~500 in: joined at a ~$50M company. Held to $1.75T = a 17,000x. The core of the club — maybe 150–250 left holding - September 2008, SpaceX has first successful launch - 2010–2016: joined at $1B–$10B. Needs a senior grant — directors, principal engineers, early Starlink. ~100–200 - C-suite board: Shotwell, Johnsen past $1B. A layer of SVPs below them clears $100M on equity, not salary. ~20–40 - Post-2016: joined at $20B–$350B. To hit $100M you'd have needed ~0.4% of the company. Impossible for an employee. This is the millionaire tier — almost none reach $100M The tally: ~400–500 at $100M A few dozen above $500M A handful of billionaires past Musk Same building. Same mission. Two orders of magnitude apart — set entirely by what year you walked in. Early isn't a strategy. It's a date stamp.

what happens to trading if AI keeps getting better? Dwarkesh Patel asked Jane Street's head of technology the obvious question: if we get AGI, can't it just replace what you do? his answer: "trading feels to me like AGI-complete." "all of the different problems of the world end up influencing what you're doing in a trading context. trading involves figuring out what things are worth. which means making predictions about the future. and lots of different things flow into that." in other words, to fully automate trading you'd need to fully automate understanding the entire world. every geopolitical event, every supply chain shift, every behavioral pattern, every regulatory change, it all feeds into price. and even at a firm running tens of thousands of GPUs with sub-100-nanosecond execution: "many of your most profitable days happen when weird stuff happens and nobody knows what's going on. doing that well involves human judgment. we think humans work better than models through phase transitions." his conclusion: "i have never been more desperate to hire more engineers and traders than i am today. everything people are doing is more valuable than it was." so what does this mean for the rest of us? if the most automated trading firm on earth is saying humans matter more not less. the implication is that AI doesn't shrink the opportunity in trading. it raises the bar for what each person can do. the traders who learn to use these tools will pull further ahead. the ones who don't will fall further behind.