Tony Kam
1 followers · 2 following · 1330 views
on the atlas — 47
- Verifiability | karpathy3 savers
- A Cynic’s Guide To Fintech. Several business models that are bound… | by Dan Davies | Bull Market | Medium2 savers
- Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's Blog1 savers
- 2025: The year in LLMs4 savers
- 2025 LLM Year in Review | karpathy16 savers
- The space of minds | karpathy2 savers
- Animals vs Ghosts | karpathy7 savers
- Chairman's Letter - 19871 savers
- Chairman's Letter - 19834 savers
- Chairman's Letter - 19791 savers
- AI Market Clarity - by Elad Gil - Elad Blog5 savers
- The SaaS Metrics That Matter2 savers
- Startup = Growth3 savers
- Enterprises vs SMBs - Bottom Up by David Sacks2 savers
- The Wilderness Period - Bottom Up by David Sacks2 savers
- The One Who Defines the Category Wins the Category2 savers
- Home | Substack1 savers
- The Pipeline Metrics That Matter2 savers
- The Sharp Startup: When PayPal Found Product-Market Fit2 savers
- The Difficulty Ratio - by David Sacks and Brian Murray1 savers
- Product First - Bottom Up by David Sacks1 savers
- The Agentic Web and Original Sin – Stratechery by Ben Thompson3 savers
- Why We Think | Lil'Log25 savers
- Language Models in Plato's Cave - by Sergey Levine4 savers
- Sporks of AGI - by Sergey Levine - Learning and Control8 savers
- Market Ending Moves - by Elad Gil - Elad Blog4 savers
- Idea Generation - Sam Altman13 savers
- Hard Startups - Sam Altman9 savers
- Content and Community – Stratechery by Ben Thompson2 savers
- Chairman's Letter - 19911 savers
- Curius / Onboarding2621 savers
- reflections on palantir - Nabeel S. Qureshi35 savers
- Reflections on OpenAI33 savers
- Startup Advice - Sam Altman17 savers
- The Only Important Technology Is The Internet — Kevin Lu17 savers
- Asymmetry of verification and verifier’s law — Jason Wei15 savers
- The Moat of Low Status12 savers
- How to Get Startup Ideas12 savers
- LLM Daydreaming · Gwern.net12 savers
- Pete Koomen12 savers
- The only thing that matters9 savers
- Why Software Is Eating the World | Andreessen Horowitz9 savers
- DOGE Days8 savers
- Life lessons from reinforcement learning — Jason Wei7 savers
- Why GitHub Actually Won6 savers
- Rippling and the return of ambition - by John Luttig2 savers
- Why Smart People Have Bad Ideas2 savers
highlights — 264
And it turns out that price-sensitive customers are often significantly worse risks; there’s often a reason why someone’s so keen on getting the cheapest deal, and it’s usually a reason that isn’t in the database you bought.
A Cynic’s Guide To Fintech. Several business models that are bound… | by Dan Davies | Bull Market | MediumSurprisingly, Nano Banana is terrible at style transfer even with prompt engineering shenanigans, which is not the case with any other modern image editing model. I suspect that the autoregressive properties that allow Nano Banana’s excellent text editing make it too resistant to changing styles. That said, creating a new image in the style of Studio Ghibli does in fact work as expected, and creating a new image using the character provided in the input image with the specified style (as opposed to a style transfer) has occasional success.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogThat’s honestly better than expected, and the prompt cost 916 tokens. It got the overall layout and colors correct: the issues are more in the text typography, leaked classes/styles/JavaScript variables, and the sidebar:body ratio. No, there’s no practical use for having a generative AI render a webpage, but it’s a fun demo.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's Blogmy guess is that this rule is a hack intended to avoid the perception of model collapse by avoiding the generation of 2022-era AI images which would be annotated with those buzzwords.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogNano Banana is still bad at rendering text perfectly/without typos as most image generation models. However, the expanded text is peculiar: it does follow from the prompt, although “Blue Blur” is a nickname for the normal Sonic the Hedgehog. How does an image generating model generate logical text unprompted anyways?
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogOne of the most compelling-but-underdiscussed use cases of modern image generation models is being able to put the subject of an input image into another scene. For open-weights image generation models, it’s possible to “train” the models to learn a specific subject or person even if they are not notable enough to be in the original training dataset using a technique such as finetuning the model with a LoRA using only a few sample images of your desired subject. Training a LoRA is not only very computationally intensive/expensive, but it also requires care and precision and is not guaranteed t…
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogWorking with the Gemini API is a pain and requires annoying image encoding/decoding boilerplate, so I wrote and open-sourced a Python package: gemimg, a lightweight wrapper around Gemini API’s Nano Banana endpoint that lets you generate images with a simple prompt, in addition to handling cases such as image input along with text prompts.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogFor developers who want to build apps that programmatically generate images from Nano Banana, Google offers the gemini-2.5-flash-image endpoint on the Gemini API. Each image generated costs roughly $0.04/image for a 1 megapixel image (e.g. 1024x1024 if a 1:1 square): on par with most modern popular diffusion models despite being autoregressive, and much cheaper than gpt-image-1’s $0.17/image.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogWhile most image generation models are diffusion-based to reduce the amount of compute needed to train and generate from such models, gpt-image-1 works by generating tokens in the same way that ChatGPT generates the next token, then decoding them into an image.
Nano Banana can be prompt engineered for extremely nuanced AI image generation | Max Woolf's BlogContext rot, by Workaccount2 on Hacker News, for the thing where model output quality falls as the context grows longer during a session.
2025: The year in LLMsThe problem is that the big cloud models got better too—including those open weight models that, while freely available, were far too large (100B+) to run on my laptop. Coding agents changed everything for me. Systems like Claude Code need more than a great model—they need a reasoning model that can perform reliable tool calling invocations dozens if not hundreds of times over a constantly expanding context window. I have yet to try a local model that handles Bash tool calls reliably enough for me to trust that model to operate a coding agent on my device.
2025: The year in LLMsI remain deeply concerned about the safety implications of these new tools. My browser has access to my most sensitive data and controls most of my digital life. A prompt injection attack against a browsing agent that can exfiltrate or modify that data is a terrifying prospect.
2025: The year in LLMsThen in November Anthropic published Code execution with MCP: Building more efficient agents—describing a way to have coding agents generate code to call MCPs in a way that avoided much of the context overhead from the original specification.
2025: The year in LLMsAnthropic themselves appeared to acknowledge this later in the year with their release of the brilliant Skills mechanism—see my October post Claude Skills are awesome, maybe a bigger deal than MCP. MCP involves web servers and complex JSON payloads. A Skill is a Markdown file in a folder, optionally accompanied by some executable scripts.
2025: The year in LLMsSince leaning heavily into Claude Code and friends myself I’ve hardly used MCP at all—I’ve found CLI tools like gh and libraries like Playwright to be better alternatives to the GitHub and Playwright MCPs.
2025: The year in LLMsThe reason I think MCP may be a one-year wonder is the stratospheric growth of coding agents. It appears that the best possible tool for any situation is Bash—if your agent can run arbitrary shell commands, it can do anything that can be done by typing commands into a terminal.
2025: The year in LLMsMCP’s release coincided with the models finally getting good and reliable at tool-calling, to the point that a lot of people appear to have confused MCP support as a pre-requisite for a model to use tools.
2025: The year in LLMsThey also shipped Gemini CLI (their open source command-line coding agent, since forked by Qwen for Qwen Code), Jules (their asynchronous coding agent), constant improvements to AI Studio, the Nano Banana image models, Veo 3 for video generation, the promising Gemma 3 family of open weight models and a stream of smaller features.
2025: The year in LLMsGemini 2.5 and then Gemini 3.0—each model family supporting audio/video/image/text input of 1,000,000+ tokens, priced competitively and proving more capable than the last.
2025: The year in LLMsinternal knowledge and token-based reasoning capabilities.
2025: The year in LLMsIn November Google fully embraced the “Nano Banana” name with the release of Nano Banana Pro. This one doesn’t just generate text, it can output genuinely useful detailed infographics and other text and information-heavy images. It’s now a professional-grade tool.
2025: The year in LLMsNano Banana caught people’s attention because it could generate useful text! It was also clearly the best model at following image editing instructions.
2025: The year in LLMsAnyone who’s tried running their agent with automatic confirmation (aka YOLO mode—Codex CLI even aliases --dangerously-bypass-approvals-and-sandbox to --yolo) has experienced the trade-off: using an agent without the safety wheels feels like a completely different product. A big benefit of asynchronous coding agents like Claude Code for web and Codex Cloud is that they can run in YOLO mode by default, since there’s no personal computer to damage.
2025: The year in LLMsChatGPT Code Interpreter in early 2023—a system baked into ChatGPT that allowed it to run Python code in a Kubernetes sandbox.
2025: The year in LLMsThe major labs all put out their own CLI coding agents in 2025 Claude Code Codex CLI Gemini CLI Qwen Code Mistral Vibe Vendor-independent options include GitHub Copilot CLI, Amp, OpenCode, OpenHands CLI, and Pi. IDEs such as Zed, VS Code and Cursor invested a lot of effort in coding agent integration as well.
2025: The year in LLMsBut if you define agents as LLM systems that can perform useful work via tool calls over multiple steps then agents are here and they are proving to be extraordinarily useful. The two breakout categories for agents have been for coding and for search.
2025: The year in LLMsBut if you define agents as LLM systems that can perform useful work via tool calls over multiple steps then agents are here and they are proving to be extraordinarily useful. The two breakout categories for agents have been for coding and for search.
2025: The year in LLMsBut if you define agents as LLM systems that can perform useful work via tool calls over multiple steps then agents are here and they are proving to be extraordinarily useful. The two breakout categories for agents have been for coding and for search.
2025: The year in LLMsBut who is actually going to build the LLM GUI? In this world view, nano banana is a first early hint of what that might look like. And importantly, one notable aspect of it is that it's not just about the image generation itself, it's about the joint capability coming from text generation, image generation and world knowledge, all tangled up in the model weights.
2025 LLM Year in Review | karpathyWe're going to see equivalents of personal computing, of microcontrollers (cognitive core), or internet (of agents)
2025 LLM Year in Review | karpathyGoogle Gemini Nano banana is one of the most incredible, paradigm-shifting models of 2025. In my world view, LLMs are the next major computing paradigm similar to computers of the 1970s, 80s, etc
2025 LLM Year in Review | karpathyVibe coding will terraform software and alter job descriptions.
2025 LLM Year in Review | karpathyNote that the primary distinction that matters is not about where the "AI ops" happen to run (in the cloud, locally or whatever), but about everything else - the already-existing and booted up computer, its installation, context, data, secrets, configuration, and the low-latency interaction.
2025 LLM Year in Review | karpathyAnd while agent swarms running in the cloud feels like the "AGI endgame", we live in an intermediate and slow enough takeoff world of jagged capabilities that it makes more sense to run the agents directly on the developer's computer.
2025 LLM Year in Review | karpathyClaude Code (CC) emerged as the first convincing demonstration of what an LLM Agent looks like - something that in a loopy way strings together tool use and reasoning for extended problem solving. In addition, CC is notable to me in that it runs on your computer and with your private environment, data and context.
2025 LLM Year in Review | karpathyPersonally I suspect that LLM labs will trend to graduate the generally capable college student, but LLM apps will organize, finetune and actually animate teams of them into deployed professionals in specific verticals by supplying private data, sensors and actuators and feedback loops.
2025 LLM Year in Review | karpathyLLM apps like Cursor bundle and orchestrate LLM calls for specific verticals: They do the "context engineering" They orchestrate multiple LLM calls under the hood strung into increasingly more complex DAGs, carefully balancing performance and cost tradeoffs. They provide an application-specific GUI for the human in the loop They offer an "autonomy slider"
2025 LLM Year in Review | karpathywe got a whole new knob (and and associated scaling law) to control capability as a function of test time compute by generating longer reasoning traces and increasing "thinking time"
2025 LLM Year in Review | karpathy"shape" of LLM intelligence in a more intuitive sense. We're not "evolving/growing animals", we are "summoning ghosts"
2025 LLM Year in Review | karpathyPeople who build good internal models of this new intelligent entity will be better equipped to reason about it today and predict features of it in the future.
The space of minds | karpathyThe more a task/job is verifiable, the more amenable it is to automation in the new programming paradigm. If it is not verifiable, it has to fall out from neural net magic of generalization fingers crossed, or via weaker means like imitation. This is what's driving the "jagged" frontier of progress in LLMs.
Verifiability | karpathyIn this new programming paradigm then, the new most predictive feature to look at is verifiability. If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well.
Verifiability | karpathyA lot of recent work is also very interested in memory (think CLAUDE.md files) as a mechanism for test-time learning that uses the text/context as the substrate instead of weights.
Animals vs Ghosts | karpathyLLMs do have their own continual learning at test time, it's just not based on weight training, but I think Sutton didn't fully react to that
Animals vs Ghosts | karpathyAnimals do observe demonstrations, but they are not strictly speaking directly supervised with actions, like supervised learning does. Animals are never teleoperated in training mode. The closest thing I can think of is if you for example help a child eat with a spoon or something, by literally holding their hand and showing the motion. Even then, it's not clear that their brains are literally training on that.
Animals vs Ghosts | karpathyIt's possible that ghosts:animals :: planes:birds.
Animals vs Ghosts | karpathyStated plainly, today's frontier LLM research is not about building animals. It is about summoning ghosts. You can think of ghosts as a fundamentally different kind of point in the space of possible intelligences. They are muddled by humanity. Thoroughly engineered by it. They are these imperfect replicas, a kind of statistical distillation of humanity's documents with some sprinkle on top. They are not platonically bitter lesson pilled, but they are perhaps "practically" bitter lesson pilled, at least compared to a lot of what came before.
Animals vs Ghosts | karpathyTLDR: Pretraining is our crappy evolution. It is one candidate solution to the cold start problem, to be followed later by finetuning on tasks that look more correct
Animals vs Ghosts | karpathysupervised learning that is ~absent in the animal kingdom
Animals vs Ghosts | karpathyIn Sutton's world view, all there is is an interaction with a world via reinforcement learning, where the reward functions are partially environment specific, but also intrinsically motivated, e.g. "fun", "curiosity", and related to the quality of the prediction in your world model
Animals vs Ghosts | karpathy