ScaffoldBench

RunOpus 5.5repeat 2

Propose-select on Rebook onto the earliest available flight and settle the difference

claude-opus-5-5__propose-select__airline-medium__r2

Score

1.000

passed
design ran as designed
Cost
$0.339
Latency
1m 2s
setup 0.0s, grading 0.0s
Turns
37
Tool calls
31
Subagents
3
Tokens in / out
123k / 8.2k
95,511 read from cache

Did the design run?

every check held
  • Run completed: heldharness.run-completedThe environment was built, the architecture returned and grading finished, none of them throwing.
  • No API failures: heldharness.apiNo model call failed at the API.
  • Every call ended: heldharness.calls-endedEvery call ended with a result, or was stopped by the run's own limits.
  • No harness tool denied: heldharness.harness-toolsThe sandbox guard never denied a harness tool, nor a tool the session itself offered.
  • Tool servers connected: heldharness.mcp-connectedEvery tool server the harness gave a call was connected when the session started.
  • No foreign tool servers: heldharness.mcp-hermeticNo call loaded a tool server the harness did not give it.
  • Builtin tools: heldharness.builtin-toolsA check recorded with the run.
  • Structured replies arrived: heldharness.structured-outputEvery call that asked for structured output and ran to the end got it.
  • No silent fallbacks: heldharness.not-degradedNo phase fell back on a default (an empty plan, a default winner) instead of its output.
  • All three proposers ran: heldpropose-select.proposalsThree proposers ran in parallel, each from its own angle, and none was starved while another ran.
  • The selector chose: heldpropose-select.selectionThe selector picked a proposal by a valid number, read from its schema reply and not a default.
  • Each call was shown the right tools: heldpropose-select.toolsEach proposer was shown the read-only tools, the selector none, and the executor the full toolset.

Phases

Proposal 1 (policy-first): 0.0s to 17.8s, finished, 7 tool calls, $0.059
Proposal 2 (state-first): 0.0s to 20.5s, finished, 7 tool calls, $0.061
Proposal 3 (outcome-first): 0.0s to 18.9s, finished, 6 tool calls, $0.059
Select: 20.5s to 36.5s, finished, 0 tool calls, $0.065
Execute: 36.5s to 1m 2s, finished, 11 tool calls, $0.094

Drawn from the phase records: timing and cost per call, no tool calls or messages.

PhaseDurationCostTurnsToolsModel
Proposal 1 (policy-first)17.8s$0.05987claude-opus-5-5
Proposal 2 (state-first)20.5s$0.06187claude-opus-5-5
Proposal 3 (outcome-first)18.9s$0.05976claude-opus-5-5
Select16.0s$0.06520claude-opus-5-5
Execute25.3s$0.0941211claude-opus-5-5

The call-by-call replay is not published for this run. Traces are large, so the repository keeps 9 of this run set’s 900 and leaves the rest out; the score, phases and final answer on this page come from the committed run record.

Runs you can replay

Record this cell yourself: npm run bench -- run --strategies propose-select --tasks airline-medium --models claude-opus-5-5 --repeats 1

Score breakdown

  • passReservation moved to the earliest flight with an available seat, same cabin
    1.00weight 0.43

    booked on AA204, expected AA204

  • passExactly one fee, the fare difference for the flight actually booked
    1.00weight 0.43

    fare difference owed on the booked flight: 8000 cents

  • passExactly one economy seat moved from the old flight to the new one
    1.00weight 0.14

Penalties

No penalties were applied.

Scorer metadata

tool calls
31
failed tool calls
0
distinct tools
9

Final answer

I've moved Dana Whitfield (R2087) to AA204, departing SFO at 09:05 on 2026-08-03, and charged her $80.00 for the change. She stays in economy, the cabin she booked.

Why AA204: it's the earliest flight that day with an economy seat.

  • UA88 at 06:15 has no economy seats left.
  • AA918 at 07:30 is cancelled. It still shows 6 economy seats, but it isn't flying.
  • AA204 at 09:05 is scheduled and had 4 economy seats at $390.00 when I checked it again just before booking.

What she pays:

  • Fare difference: $80.00. That's the $390.00 fare minus the $310.00 she paid. The economy fare rules collect the full difference whatever the passenger's tier, so it's recorded as a single fare-difference charge.
  • Change fee: none. It's normally $75.00, but she's a Gold member and Gold members don't pay it.

Nothing else changed. There's no refund or cancellation, her one checked bag is as it was, and I didn't touch any other booking, passenger or flight. The airline system's date is 2026-07-28, so the flight is still in the future.

What to tell Dana: "You're now on AA204 at 09:05, the earliest flight that day with an economy seat. The 06:15 UA88 has no economy seats left and the 07:30 AA918 is cancelled. As a Gold member you don't pay the $75 change fee, but you do owe the $80 difference between the new $390 fare and the $310 you paid. Everyone pays the fare difference, whatever their status."

Provenance of the numbers on this page

run set
scaffold-or-scale-v1
models
Haiku 4.5, Sonnet 5.5, Opus 5.5
effort
high
repeats
2
fixtures
replay
runs
900 / 900
strategy cost
$354.44
judge cost
$9.19
wall clock
9h 48m
git
fce5807b2d

Cost excludes judge calls, which are tracked separately so a scoring model never flatters or penalises an architecture. Every figure on this site is read from results/ at build time.