Six frontier models hit our partner nodes this month. Hard-wire to one vendor and you rebuild every quarter. We bet on the opposite. Build the graph, swap the models underneath it. A workflow in a JSON file outlives every model inside it. That bet: → 4M+ users, 50K downloads a day → 62K GitHub stars → 106K by March → NVIDIA shipped NVFP4 support with us at GDC: 2.5x faster generation, 60% less VRAM on RTX 50-series → Silverside AI built SVEDKA's Super Bowl spot - the first primarily AI-generated Super Bowl ad on ComfyUI → Magnopus published its architecture for running ComfyUI on shared production infrastructure → "ComfyUI artist" is now a job title on studio postings Recruiters are a lagging indicator. When they move, the argument is over.

Intelligence is transitioning from scarcity to abundance. ( @cpaik went deeper on this in his latest essay: .) We believe in choice and democratic access to intelligence. We think that the number of models in the future will be closer to the number of species, rather than one or two gatekeeping monopolies. We invested in @OpenRouter for exactly this: a neutral layer that brings order and liquidity to the chaos of this Cambrian explosion. Yesterday, OpenRouter took its next step, alongside @Stripe. Stripe built much of the connective tissue of the internet economy, and they recognize the same shape in what OpenRouter is building for intelligence. Credit, above all, to @alexatallah. Stripe is fortunate to be working with a founder this talented. Twice now he has built the exchange that gives an emergent ecosystem its liquidity — first for digital goods, now for intelligence. The open future, where access is abundant and people have the most choice is one that we are excited about. We remain committed to helping OpenRouter, and now Stripe in realising this future.

Alex is a generational founder and it’s been an honor to work with him at @PaceCap. OpenRouter has stewarded and built towards a future of model diversity and democratized access; a future we deeply believe in. Congratulations to everyone at @OpenRouter and @stripe!

Comfy MCP is now local and open-source! The #1# ask after Comfy Cloud MCP shipped in June. Your agent sees your actual install. So every node, custom nodes included, and every model on disk. It fetches files, starts your instance, gets a workflow to the point where it runs. It is now the easiest way to help with your local Minimax H3 workflows! Cloud MCP still does everything it did. Tell your agent where a job goes, or let it decide.

OpenRouter is one of the most insane stories in tech. Co-founded by OpenSea founder, Alex Atallah. It has become one of the most important companies in AI. Scaled to 100s of millions in revenue and wildly profitable. They process 25 TRILLION tokens every week and will do 1 QUADRILLION tokens this year. As a result, Stripe have reportedly offered to buy them for $10BN. Last round was $1.3BN just months ago… so what happens now? I sat down for a chat with OpenRouter Founder, @alexatallah and have summarized my notes below: 1. This Is Going to Be the Biggest Market in the History of Tech AI model inference is set to become one of the largest markets in technology. Even as token prices fall dramatically, Jevons Paradox means usage can grow far faster than costs decline, driving massive overall compute consumption across an increasingly multi-model future. 2. Why None of the Routing Products Being Created Today Will Compete With OpenRouter Building AI gateways has become trendy, but copycat routers that treat routing as a side feature are playing to exist rather than to win. True routing requires relentless focus on optimizing latency, cost, and constantly shifting model quality while giving developers maximum flexibility. 3. Model Labs Have Every Incentive to Come After Your Startup Eventually Startups building thin wrappers around AI models face an existential threat if they occupy workflows that frontier labs consider strategic. Releases like Claude Design show how labs can move up the application layer, absorb valuable use cases, and lock entire enterprise teams into their ecosystems. 4. One New Model Every 10 Hours and the Myth of Model Consolidation Model creation is accelerating as hardware companies, Neo Labs, and agent frameworks continuously release specialized models. Rather than consolidating around a single winner, developers are embracing diverse models optimized for different tasks, reinforcing a fragmented, multi-model future. 5. America Is Still Very, Very Behind in Open-Weight AI The U.S. remains significantly behind China in the race for open-weight models. Chinese labs benefit from aggressive state support and open-source momentum, while American efforts face business model and capital constraints. Closing the gap will require dedicated access to compute and sustained investment. 6. How Open-Weight Models Are Rapidly Closing the Gap on Frontier AI Models like Kimi K3 and the 5.2 generation show how quickly open-weight intelligence is closing the gap with proprietary frontier models. Using low-cost open models for deterministic subtasks beneath a frontier orchestrator can maintain output quality while sharply reducing inference costs. 7. Why We Need to Think About Employee Cost Completely Differently In the AI era, employee cost is no longer just a fixed salary. It increasingly includes the variable inference spend of the AI tools each employee deploys. Companies will need to evaluate workforce performance through a cost-to-productivity lens that accounts for both human compensation and AI consumption. (links in comments)

Josh's insight here is unique because he's had the best seat in the house for years. The browser is both a utility and a leisure tool, and the team at @browsercompany has seen exactly how people use it. Consumers want their leisure amplified. They don't want productivity! Productivity has to be found, and finding it is *work*. Education is work. DIY is work. And an agent asking you a question is really the agent handing you a microtask. Very few people want to manage and even fewer want to be interrupted in order to do it. It's the same reason notifications have such a narrow band where they help before they annoy. Meanwhile the For You page is the most successful consumer agent ever built and yet no one calls it one. It runs inference on your behalf constantly, makes thousands of decisions a session, and asks you for absolutely nothing. That's where consumer is headed. Companies using AI to spoonfeed people more of the leisure they already want, not doling out microtasks.

We are making the updated DeepSeek V4-Flash 0731 free in Cline. This is the first flash model we've found performs at SOTA levels, and are excited for you to feel the new frontier. 1. npm i -g cline 2. Open /settings > Cline provider 3. Select deepseek-v4-flash

We've been asked why Cline works better with open weights models like DeepSeek than other harnesses and want to share some detail (more in our blog below). Our internal evals show that many open weights models like DeepSeek, GLM, and Kimi are RL trained to spend more tokens verifying their work: run tests, check the build, re-read diffs before completing. We believe this is how they were able to close the gap with closed model performance. Meanwhile closed models and their harnesses tend to optimize for efficiency and fewer tokens (which makes sense when massively subsidizing subscription plans). So we've been intentional in building the Cline harness to let the model work however it was trained to work. At open weights pricing, the baked-in paranoia burning extra tokens verifying work gets significantly better results for less money. In our benchmark runs, we've seen ~20% gains over competing harnesses just from changes that let the model lean into its RL training.

DeepSeek silently updated their changelog with a new V4-Flash upgrade 1 hour ago. Their new Terminal-Bench score is 82.7, a massive +25.8 point leap from its initial April preview score of 56.9. Currently only available via their API, open weights release will follow shortly.

Dialectic 52: @jackconte! Jack is not an entrepreneur who makes art on the side. He is a lifelong artist and musician who became the unlikely founder and CEO of @patreon, the only company he has ever worked for. At his (musical) peak, Jack was releasing 100 music videos a year between his bands @pomplamoose and @scarypockets while building one of the few enduring creator platforms on the web. He started Patreon to solve his own problem as a full-time YouTuber: he wanted fans to pay him directly for his work rather than accepting a dashboard that told him 3 million views were worth $18. Patreon is rare amongst its peers: it's a platform built first and foremost for creators, not consumers. It makes money when creators get paid. Jack believes that incentive alignment has helped Patreon earn creators’ long-term trust, and will help it endure as mediums, technology, and platforms evolve. I talked to Jack about all of this and more: - creative endurance and becoming “post-chaos” - why Silicon Valley undervalues art, brand, and storytelling - why art isn't content and getting paid isn’t selling out - creator-first incentives, trust, and 100 years of Patreon - why Jack is still the right CEO after 13 years - leadership: ambition, obsession, self-reflection, and learning from people better than you - Charlie Kaufman, authenticity, and making fans - lucid dreaming and taking more risks Timestamps: (0:00) - Opening Highlights (1:38) - Intro to Jack & Thanks to Notion (3:44) - Start: Prolific Output, Energy, Juggling, Endurance, and Becoming Post-Chaos (22:04) - Combining Creativity and Business and Art vs. Content (33:21) - Advertising vs. Membership & Patronage vs. Patreon (44:06) - Jack’s First and Only Job, and Silicon Valley’s Undervaluing of Creativity, Brand, and Storytelling (and Patreon) (55:17) - Building for Creators First, Incentive Alignment, Trust, 100 Years of Patreon, and Why Jack Is the Right CEO for Now (1:04:49) - Leadership: Types of Ambition, Obsession, Self-Reflection, Bias to Action, and Learning from Others (1:12:59) - Talent vs. Enthusiasm, Being Proud of What You Make, Making Fans and Being True, Managing Parasociality, and Community (1:28:30) - Filmmaking, Labels and Collectives vs. Individuals, Breaking Patterns, Creative Intimacy, Music, Patreon’s Origin, Persistence, Lucid Dreaming, and Taking Risks (1:49:26) - Thanks Again to Notion @dialecticpod Ep. 52: Jack Conte - Draw the Next Page - is out now, below and on all platforms.

Introducing AI Game Theory: An interactive sandbox that explores power and leverage between players in the AI economy. Every CEO or investor you meet has a built-in bias shaped by their vantage point. The goal of this project is to make those incentives more legible.