
Early next week, we are transferring 140M OP Tokens between wallets for operational purposes. This is an internal, planned transaction and we're sharing a heads up to keep our community informed.

The official Optimism Governance account.

Early next week, we are transferring 140M OP Tokens between wallets for operational purposes. This is an internal, planned transaction and we're sharing a heads up to keep our community informed.


Submission for Optimism Growth Grants are now open! Our Objective: Grow DEX TVL and increase trading volume. Apply today 👇


Later today, we are transferring 6,400 ETH between wallets. This procedure enables the liquid staking of the Optimism Collective's ETH treasury, you can read more about it below 👇



Calling on all Optimism Citizens & Delegates 🏛️ You have 25 hours left to vote on the Proposal to Align OP Token with Superchain Success. Vote now to shape the future of the Superchain!

Tomorrow, we are transferring 20M OP between wallets for operational purposes. This is an internal, planned transaction and we’re sharing a heads up to keep our community informed.

Later today, we are transferring 30M OP between wallets for operational purposes. This is an internal, planned transaction and we’re sharing a heads up to keep our community informed.

Tomorrow, we are transferring ~15M OP between wallets for operational purposes. This is an internal, planned transaction and we’re sharing a heads up to keep our community informed.

fin/ In the future, we’re confident that AI agents will play an important role in minimizing the time and effort required by humans to participate thoughtfully in governance. Vitalik’s principle of "AI as the engine, humans as the steering wheel” rings true in our findings from this experiment.

5/ Well-established decision making frameworks – like the Law of Chains – are necessary to help AI agents reason about proposals but are insufficient alone. A well-defined set of value preferences is also necessary to guide an agents reasoning in instances where trade-offs are necessary, and this is where human input is needed.

4/ In the synthetic proposals that we tested agents against, we observed that the biggest difference in reasoning came down to how agents interpreted the Law of Chains. While some agents saw it as inviolable set of rules, others assigned less importance to it or interpreted it more flexibly.

3/ We didn’t observe meaningful differentiation in the reasoning capabilities of these custom architectures when compared against off-the-shelf models like GPT-5 Deep Research. However, we did observe a meaningful reduction in cost from these custom architectures, since many were capable of using lower-cost models.

2/ We sponsored teams at @chain_haya, @x23_ai, @fission_web3, and @elizaosteams to develop custom agents using state-of-the-art models and custom architectures like mixture-of-agents and mixture-of-experts. We then tested these agents, as well as off-the-shelf state-of-the-art reasoning models, in a series of synthetic governance proposals.