What Osaka changed, transaction by transaction
fevm implements Osaka, the fork Ethereum took on 3 December 2025. Replaying blocks from the weeks before it, under Osaka's rules, asks a question nobody can ask a live node: which real transactions would have ended differently under the new rules, and why?
| Block | Transaction | Cause | On chain (Prague) | Under Osaka |
|---|
How it was measured
Each transaction ran on its own pre-state, fetched from a node with the prestate tracer, and its outcome was compared with the chain's receipt and state diff, as for every replay. Under Osaka rules a difference is expected only where a rule changed. For each one, fevm-replay traced the transaction's calls and checked which changed rule it touches: a gas limit above 224, a call to the MODEXP precompile (0x05), or a call to 0x100. Every difference has exactly one of these causes; none is unexplained.
The rest of the transactions in these blocks match the chain exactly, so they behave the same under both forks. Blocks 23,700,000 to 23,930,009, sixty in all.