Token costs are why there will be no saas apocalypse / good dev tools are cached intelligence for agents! The popular theory goes: agents can write code, so they'll just rebuild every tool from scratch and hit raw APIs. no more dev tools, no more CLIs, no more software layers. just agents and endpoints! We just tested this and the data says the opposite. We benchmarked Claude Code and Codex on real Hugging Face Hub tasks (~1,000 graded runs), with two setups: the agent-optimized hf CLI vs the agent hand-rolling curl or SDK calls from scratch. Hand-rolling burns up to 6x more tokens on multi-step tasks and fails more often (84% vs 94% task success). And that's just dropping one abstraction layer. It would obviously be orders of magnitude more tokens and a dramatically higher failure rate if the agent tried to bypass HF altogether and rebuild model hosting, versioning, and distribution from scratch. Every time an agent re-derives a workflow from raw API calls, you pay for that reasoning in tokens. every single run. a good CLI compresses that entire chain into a few high-level commands the agent can't get wrong. In a world where everyone is complaining tokens are too expensive, abstraction is leverage: thousands of hours of design decisions your agent doesn't have to re-reason about at inference time. Good tools are cached intelligence for agents! So no, agents won't rebuild everything from scratch. they'll gravitate to the most token-efficient tools, because that's what their owners pay for. The software that survives won't just be accessible to agents, it will be accurate and cheap for them to drive. We're seeing it happen with HF, which is becoming the platform for agents to use AI: ~49M requests in just two months, and growing fast! huggingface.co/blog/hf-cli-f…

the cost of shipping code went to zero taste didn't but "taste" sounds mystical and unfixable, so nobody teaches it. here's the unmystical version: taste is just an eval you haven't written down yet how you choose what to measure is what matters 1/8

It's time to fly.

Feels quite magical to be able to clone a 68 TB dataset to my private HF training bucket while I only have a 4TB local disk, all of that in less than a minute thanks to HF infra optimizations & xet dedup!

We are also releasing an interactive UMAP visualization tool, which allows users to navigate the dataset and a retrieval space to extract specific sub-distributions. 🤗 UMAP: huggingface.co/spaces/jasper… 🤗 Retrieval space: huggingface.co/spaces/jasper…

📢 New @heyjasper release ! 📢 MONET 🌸 : An Apache2.0 deduped and recaptioned dataset of 105M samples unlocking reproducible text-to-image research. Nano T2I 🖌️ : A codebase to train your own T2I model 🤗 @huggingface: huggingface.co/datasets/jasp… 💻: github.com/gojasper/nano-t2i Very excited about this new release, pushing the boundaries of open and reproducible T2I research. Congrats to the team! Benjamin Aubin Gonzalo Quintana @onurxtasar @UlaLaParis @_jeev2 @dh7net @clipdropapp @heyjasperai

The argument for having ownership over your own data is so much easier when you focus on the utility you gain rather than ideology. Reasoning over your own data should be permissionless 🦅

I'm late to the party, but cmux is great. github.com/manaflow-ai/cmux current split: codex mac app: knowledege work, learning, reading cmux codex cli: coding

Spinning Mt Fuji Like A Beyblade 🤯