Run comparison

What changed between two runs — and what didn't

Runs are compared only when pack, network and config match. When the adapter differs, raw consistency and app results are shown as separate diffs and no winner is named

Run Arun_1e7c94a05 Nov 2026, 11:44 UTC · 181 requests
→
Run Brun_6dc49f727 Nov 2026, 15:06 UTC · 186 requests
ComparabilityRun ARun B
Fixture packapp-mainnet-core@3app-mainnet-core@3match
Pack sha256b73d17280316f4d7b73d17280316f4d7match
Network / genesismainnet-beta · 5eykt4…2N9dmainnet-beta · 5eykt4…2N9dmatch
Profile & capsmainnet-prod · finalized · 300 reqmainnet-prod · finalized · 300 reqmatch
Endpointsprod-primary + public-refprod-primary + public-refmatch
Adapterapp-decoder 2.3.1app-decoder 2.4.0differs
Target version (observed)solana-core 4.4.0solana-core 4.4.1info only
0raw changes
2app changes
13unchanged

Raw responses are byte-identical to Run A on every comparable fixture. Two failed Jupiter routes now decode with the error on the wrong instruction — the change sits in the adapter bump 2.3.1 → 2.4.0. The provider's move to 4.4.1 is recorded and does not, by itself, explain anything

Per fixture · changed rows first

Stages compared: Transport, RPC capability, Data coverage, Consistency, App decode. Lamports and other moving values are never diffed