Pinned Tweet
Hot take: in a year, handing your wallet keys to an AI agent without an OnchainBench score (or with a low one) will look like reading your CVV over the phone to “Bank Support”
Today we are launching OnchainBench, the benchmark for agents in crypto.
It helps you decide how far you can trust an AI agent with your keys: how often it finishes the job, what tasks cost, and how easily it falls prey to an attack.
People are starting to let AI agents swap, lend and even trade for them. The bigger the task, the greater the risk you're taking by trusting the agent. So before you hand it your keys, you need to assess it.
There are many risks to assess at once. An agent could buy two tokens for a liquidity position but never add them to the pool, so the job isn't complete. It could spend more on AI model calls than the task is worth. And, onchain, there's no undo: agents read web pages, files and tool descriptions that anyone can write, and if one of them tricks the agent into sending funds to the wrong address, the money is gone. Until now, there was no common way to evaluate agents on all of this and more.
OnchainBench helps you compare agents by giving them the kind of work people already ask them to do: swaps on @Uniswap, @SushiSwap and @Honeyswap, lending on @aave, deposits into sDAI from @SkyEcosystem, prediction markets on @Omen_eth, research, and jobs that run over several days.
Each task runs through a @safe smart account on a copy of a public chain. The first version covers @ethereum, @gnosischain, @0xPolygon, @arbitrum, @Optimism, @base, @Celo, @modenetwork and @robinhoodapp. The benchmark emulates the kind of contracts, liquidity and prices the agent would meet with your money, without putting any of it at risk. A task only counts as done if the chain shows it was done. Some tasks also hide an attack in what the agent reads, such as a look-alike token or a fake "verification" request, so you can see how often agents get tricked into moving funds before you trust one with yours.
In the first run, Claude Code (@claudeai), Codex (@OpenAI), Hermes (@NousResearch), OpenClaw (@openclaw) and our own new OCB agent on @autonolas worked through 147 private tasks they had never seen. They completed between 88.4% and 96.2% of tasks, at about $0.03 to $0.35 per completed task. Agents lost funds to an attack 13 times in 1,793 attack tests.
We are excited to contribute this to the crypto ecosystem. If you build a protocol or a chain, agents are already among your users, and will likely be one of your biggest sources of growth. OnchainBench shows, protocol by protocol and chain by chain, how well they handle yours.
For the AI ecosystem: want an agent eval with real ground truth and real adversaries? Onchain, every result can be checked and every injected instruction is a live risk. OnchainBench is open to contributions, uses private tasks, and is ready for your agent.
Help us make onchain agents harder to trick and better at the job. If you run a protocol or are a heavy user, propose the task you think agents still get wrong on the site below, and every agent on the benchmark could end up tested on it. If you build agents, run OnchainBench yourself and submit your agent to be run.
The leaderboard is live now at onchainbench.ai
Share it with a friend!
2
9
82
Make your autonomous and manual agents work on schedule like a pro
Running several agents in Pearl can mean coming back to start each one by hand.
Auto-run is designed to start them one after another: when an agent has done its work for the day, the next one gets its turn.
Pearl Connect can now take part too. It runs for its set minutes, then hands over.
Turn on Keep Device Awake in Settings to help keep your computer from sleeping while Auto-run is on.
See how it works, then try it in Pearl: pearl.you
1
27
New state-of-art, still small open prediction model form Olas/Valory
GPT 5.6 level of intelligence, locally on your laptop!
A lean, open-weight model that comes within 1.2 points of GPT-5.6 at forecasting real-world events.
Core contributor @valoryag trained Olas-Predict-Qwen3.8-27B, an open-weight forecasting model built on Qwen3.8-27B, using reviewed forecasting examples and reinforcement learning. It follows Olas-Predict-R1-14B, Valory’s first Olas Predict model.
In Valory’s evaluation on 6,824 prediction markets created after GPT-5.6’s training cutoff, with the same prompts and evidence for both models:
• Balanced accuracy (higher is better): 67.3% vs 68.5% for GPT-5.6
• Brier error (lower is better): 0.2124 vs 0.2029 for GPT-5.6
• Against Olas-Predict-R1-14B: 22% lower Brier error
It gets this close from a starting point that scores 15.6% on the AA-Omniscience factual-knowledge benchmark, against about 59% for GPT-5.6.
Most of the gain came from the stronger starting model. Valory’s domain training added 5% lower Brier error and 12% lower calibration error on top.
Next, Valory aims to close the remaining gap in future training experiments.
Read the results: olas.network/blog/meet-olas-…
6
68
Liminal Designer retweeted
Bigger wasn’t better this time.
Core contributor @valoryag fine-tuned NVIDIA’s Nemotron 120B on forecasting data from Olas agents, using the recipe behind Olas-Predict-R1-14B: selected reasoning examples, a supervised warm-up, then reinforcement learning. Training ran through @thinkymachines’ Tinker.
On 11,496 held-out predictions, Nemotron showed no clear improvement. Before → after training:
• Brier error (lower is better): 0.265 → 0.267
• Balanced accuracy (higher is better): 59.7% → 59.4%
• Unreadable answers (fewer is better): 503 → 528
Our trained 14B model had better calibration, higher balanced accuracy and just 10 unreadable answers. Nemotron retained a slightly lower Brier error: 0.267 versus 0.270.
Why didn’t the recipe transfer? Nemotron’s mixture-of-experts design may need more varied data or different training settings for forecasting. We haven’t tested those explanations, so the cause remains open.
What we learned: our existing recipe and training budget improved the 14B model but brought no clear gain to Nemotron.
Next, we’ll test a larger dense model, with an early check for improvement before committing to the full run.
Read the experiment and results: olas.network/blog/nemotron-3…
A model for forecasting real-world events
Olas prediction trader agents generated the data. @valoryag used it to train a better AI forecaster.
Introducing Olas-Predict-R1-14B: an open-weight reasoning model with forecasting performance on par with the much larger closed-source model, GPT4.1.
My agent just made 5.1× ROI on Omen Markets! 🚀
Check it out👇
predict.olas.network/omenstr…
5
43
Liminal Designer retweeted
This week in Pearl:
→ Polystrat gets improved market selection
→ Pearl's interface is now designed to be clearer, less confusing and easier to use, with copyable text in agent chat and simpler funding information during onboarding
→ 4 users unblocked by Pearl Success Agent
Set up your first agent in minutes: pearl.you
Agent operates multiple tokens, and I can fund it with just one token on a chain I hold my funds - beautiful
Funding an agent can mean bridging, swapping and finding gas tokens on the right chain first.
In Pearl, you can fund your agent with one transfer of a token you already hold on the chain of your preference.
Choose the chain and token you want to pay with, then send the amount shown to one address in one transaction. Pearl is designed to bridge and swap it into the tokens your agent needs on its chain.
Try it in Pearl: pearl.you
Liminal Designer retweeted
$OLAS is not a cashflow token....yet.
You're betting on an incredible team of 15 - with 5 PhDs on staff.
Can they generate a lot of cashflow? Yes, but that'll come as they scale the Mech Marketplace where agents can go to purchase tools and other skills from other agents.
olas.network/agent-economies…
Liminal Designer retweeted
Live now: @david_enim on The Protocol Roundtable, discussing agent economies and more!
Watch here: youtube.com/watch?v=mlCx4rNB…
Don't trade on-chain with your hands in 2k26 - let AI do that for you
One prompt is all it took to build this autonomous AI agent for on-chain trading with Pearl Connect
Join the Olas Telegram to learn more: t.me/olaschat
5
45
It's gonna be a beefy discussion, don't miss it!
AI agents are becoming economic actors, but can we trust them?
Join @david_enim tomorrow as he explains how Olas agents are designed to handle their own wallets securely, with the owner holding the keys.
🗓 3 PM UTC
1
21
The more complex actions software can do, the easier should be the UI/UX. Transmit your intent literally and let it update on its own
Easier control over your Pearl agent: its activity and its chat, now in one place.
Agent chat has moved from the Profile to the Overview tab on the main screen, where you also see what your agent does. Get to know your agent, and adjust it by asking.
For Polystrat, Omenstrat, Optimus and Basius: olas.network/blog/chat-ui
Demo for illustration only - the product may differ - not financial advice.
Liminal Designer retweeted
Easier control over your Pearl agent: its activity and its chat, now in one place.
Agent chat has moved from the Profile to the Overview tab on the main screen, where you also see what your agent does. Get to know your agent, and adjust it by asking.
For Polystrat, Omenstrat, Optimus and Basius: olas.network/blog/chat-ui
Demo for illustration only - the product may differ - not financial advice.
You can enable your AI agents to work for you more!
Your agent, your pace: here's how it works in Pearl.
Ask your Pearl agent for its daily activity target in the chat on its profile, then give it a new number. It's designed to keep working until it gets there.
pearl.you
6
71
Liminal Designer retweeted
Your agent, your pace: here's how it works in Pearl.
Ask your Pearl agent for its daily activity target in the chat on its profile, then give it a new number. It's designed to keep working until it gets there.
pearl.you
Best marketing lead I know in crypto ✨
I hosted the largest side event at @token2049 in Singapore last year
the event got 4,500+ registrations and a record-breaking attendance, with people lining up for 3 floors at the @ArtSciMuseum just to get in
this year I got laid off, so here I am, not so subtly showing off what I can do xD
I'm looking for Marketing/Events/Growth Lead roles at start-ups. I've:
→ built and hosted 10 @AgentsUnleashD events worldwide for @autonolas across 8 cities, with the event series collecting 12,000 registrations in <1 year
→ worked with speakers and partners from @CoinbaseDev , @NEARProtocol, @safe, @SentientAGI, @Gaianet_AI, @gnosischain, etc.
→ led marketing from 0→1 at two startups
→ generated 4.7M+ organic X impressions through my marketing campaigns
if you are hiring a banger Marketing & Events lead or know someone who is, please DM me or comment👇🏻
o/w you can like share retweet repost and help make this post reach the right people. TYSM <3
portfolio ⤵
1
4
110
Liminal Designer retweeted
It feels possible to me that these fine-tuned models could begin to drive agentic demand through on-chain marketplaces.
Right now there's little benefit to agents making requests of each other, because there's little specialisation. But a proliferation of specialised models could change that.
cf @autonolas recent fine-tuned model for participating in prediction markets.
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).
I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.
We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:
"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"
I imagine we'll have some backend services that could be replaced entirely by this new architecture.
More freedom and control for our agents, thanks!
Your agent, your pace: Pearl agents can now produce more activity.
Set a daily activity target from the chat on your agent’s profile, and your agent is designed to keep working until it reaches it.
Pearl Connect, which has no staking contract, gets a daily target too, so it can now run alongside your other agents in Auto-run.
More Pearl agent activity 🤝 More Olas agent activity overall.
Read more: olas.network/blog/pace
1
7
58
Smart on-chain A2A payment solution for @autonolas agents. Recommend reading the blog 🗒️
Pearl agents just moved their spending home 🏠
- Before, price data and LLM calls used to go to outside services via x402.
- Now, Mechs on the Olas Marketplace serve them on-chain, and requests pay a protocol fee to Olas.
Same user experience. More demand on the Olas Marketplace.
Find out more: olas.network/blog/pearl-mech
2
44

















