Runta retweeted
Pi 1.0 runs on Runta today.
Each agent gets its own machine: hands-on in a browser terminal, or headless over an API with live streaming, steering, and cancel. Pi never sees your real keys.
Our egress layer swaps in model and credentials on the way out.
Runta retweeted
Full teardown here, including how Muse's Sentinel checks requests even after a command is already running: runta.com/blog/grok-bot-vs-m…
Runta retweeted
Your agent doesn't need to escape the sandbox to leak your files.
It just needs to run curl.
We looked inside Grok Bot and Muse to answer one question: what happens after the injection lands?
Our take: command review isn't enough. The check belongs at the exit, run by something the agent can't turn off.
Runta retweeted
FrontierHarness v1.1 is coming.
We're putting together an expert advisory group with maintainers of well-known coding-agent harnesses, plus related researchers from Stanford, Berkeley, and Caltech. A top-tier inference partner is also ready to sponsor the next eval runs.
In v1.1 we'll cover more models, more harnesses, and a more representative task set. Expect an order-of-magnitude jump in both run count and tokens used, so the harness × model picture gets clearer.
More soon.
Runta retweeted
This is the most important bench at the moment @runta
frontierharness.org/
Runta retweeted
Grok Bot and Muse are great. We built an open-source version and called it Errand.
Hand it a task, close the laptop, come back in the morning.
Runta retweeted
A cool thing is that you can use Runta to have exo agent play The Legend of Zelda: Link's Awakening
Try EXO on Runta under 2 mins with $50 free credit.
runta.com/blog/introducing-e…
Runta retweeted
Show HN: FrontierHarness Eval – 9 harness, same model, cost per pass varies 17x frontierharness.org/ (news.ycombinator.com/item?id…)
Runta retweeted
really cool benchmark!! and amazing work @guanlan!!
I was looking for a way to benchmark my own harness too and it was hard to quantify what's a good harness!!
I did run the experiment with graff, and here are the results!!
(would love to collab on things if you're free!)
Runta retweeted
Front page of HN, 1700+ posts on X.
Turns out a lot of people have been wondering what the harness layer actually costs , and whether the expensive ones are any better.
The harness war is on.
Runta retweeted
Remarkable result for Exo, best price / performance. Exo's design philosophy is to expose the full harness code to the model for self improvement to be maximally bitter lesson aligned. Not just the prompt, but the entire running code and logs with ability to upgrade dynamically.
Runta retweeted
My first eval project at @runta during my internship! A big takeaway from the work: harnesses matter just as much as models. Run identical tasks through different harnesses and you will see wildly different bills, pass rates, and wall-clock times.
v1.0 focused on software engineering and terminal tasks. Next we will test the full harness × model grid. Much of what we observed points to harness-model fit rather than harness quality, and we want to identify which combinations maximize pass rates while minimizing cost.
Leaderboard and the long version: frontierharness.org/
Runta retweeted
A year ago the question was which model. Now it's which harness.
Pi, Exo, Claude Code, Codex, DeepSeek Harness and 4 others. Same model, same tasks, same runtime. 360 runs, 2 billion tokens.
Pass rates: 50% to 67%.
Cost per pass: $1.05 to $18.34.
Introducing FrontierHarness Eval. 🧵
Runta retweeted
Agents aren't software. The layer beneath them has to be rebuilt. A conversation with Jeff Dean helped me see why.
Runta retweeted
Yes, we are hiring!
We are building the execution layer for next billions of agents, and we are looking for high-agency engineers:
Core Runtime (kernel-and-VMM-layer)
Cloud Infra (IaC and multi cloud)
Product Engineering (Agent Experience and DX)
Small team. High talent density. High ownership.
Apply here:
runta.com/careers/
Just posted a new list of open roles / startups hiring right now - curated by our team at @a16z !
This week featuring @guanlan (just raised $20M seed for Runta), @simon_mo_ (recently raised $150M seed for Inferact), @RussTedrake (just raised $300M seed for Walden Robotics), @maclennanm (hiring for Arc), @JackOswald10 (just came out of stealth with $80M Series A for Singularity), @noah_schochet & @NoahMcGuinn (just raised $115M for Terrafirma), @eldsjal (just raised $700M for Neko Health), @jackdent (just raised $400M for Chai Discovery), and many, many more.
Full list in thread:
Runta retweeted
Exclusive: Andreessen Horowitz backed Runta, a startup building sandboxes and guardrails to keep AI agents under control.
Read more: thein.fo/4yuP02c
Runta retweeted
Thrilled to be partner with @martin_casado.
Few people have thought harder about what happens when the infrastructure layer shifts.
a16z.com/announcement/invest…
