Skip to content

Claude Opus 5.5 on Agent bake-off: find a premium flight home

AClaude Opus 5.5 · Claude Code 2.1.299 (headless -p) · cloud · 2026-10-14 · by @bobsiopener

Trust 80%Fully autonomoussuccess · first try
Estimated cost
$0.11
Wall time
15s
Human step-ins
0
Outcome
Pass

Compare with other models on this task →

Summary

Exact prompt

Mock flight search (no real booking). flights.json lists available flights. I need a flight home to SJC or SFO from LGA or JFK on October 5th or 6th 2026, in business or first class, nonstop only, departing at or after 09:00 local. Choose the cheapest flight meeting all constraints. Write choice.json: {"best": flight_id, "runner_ups": [up to 2 next-cheapest qualifying flight_ids], "reason": "..."}.

Task prompt given to the agent (headless, offline, synthetic fixture inputs): Mock flight search (no real booking). flights.json lists available flights. I need a flight home to SJC or SFO from LGA or JFK on October 5th or 6th 2026, in business or first class, nonstop only, departing at or after 09:00 local. Choose the cheapest flight meeting all constraints. Write choice.json: {"best": flight_id, "runner_ups": [up to 2 next-cheapest qualifying flight_ids], "reason": "..."}. Inspired by https://x.com/twid/status/2101789185598247390 (unverified; not checked with the author). This is an independent re-run on a reproducible synthetic task; it does not report or reproduce the original post's numbers. Cost shown is a list-price API-equivalent estimate (run on a Claude Pro subscription, not billed).

Success criteria: All automated rubric checks pass: best = cheapest qualifying; runner-ups are next cheapest qualifying; reason given.

Setup notes: Cost is an ESTIMATE: Claude Code's list-price API-equivalent figure. The run used a Claude Pro subscription, so nothing was billed. input_tokens = all input incl. cache-creation and cache-read tokens; cached_tokens = cache reads (subset of input). Claude Code's cost bills cache writes at the 1h cache-write premium, so it can differ from a flat input-price recompute. Web disabled; tools Read/Write/Edit/Glob/Grep/Bash; single run; scored automatically (3/3 checks).

Estimated cost List-price estimate; not actually billed.

Metrics

Estimated cost
$0.11
Wall time
15s
Human step-ins
0
Turns
4
Model usage
ModelInput tokOutput tokCost
AClaude Opus 5.5anthropic/claude-opus-5.578,5771,367$0.11

Reproduce this run

  1. Use the use case prompt on the use case page.
  2. Same setup: Claude Opus 5.5 (anthropic/claude-opus-5.5) via Claude Code 2.1.299 (headless -p), cloud.
  3. It counts as a success only if it meets every success criterion: All automated rubric checks pass: best = cheapest qualifying; runner-ups are next cheapest qualifying; reason given.
  4. Log model, tokens, wall time, human step-ins and the evidence log, then submit it as a reproduction. A different model? Submit a head-to-head rerun.

How runs are scored: Methodology.

Was this useful?

⇄ Rerun on another model

Automated checks

Trust 80% comes from these checks (how Trust % works).

Evidence log
{
 "duration_api_ms": 13603,
 "stop_reason": "end_turn",
 "session_id": "[REDACTED]",
 "total_cost_usd": 0.1126462,
 "usage": {
  "input_tokens": 8,
  "cache_creation_input_tokens": 8918,
  "cache_read_input_tokens": 69651,
  "output_tokens": 1367,
  "output_tokens_details": {
   "thinking_tokens": 120
  },
  "server_tool_use": {
   "web_search_requests": 0,
   "web_fetch_requests": 0
  },
  "service_tier": "standard",
  "cache_creation": {
   "ephemeral_1h_input_tokens": 8918,
   "ephemeral_5m_input_tokens": 0
  },
  "inference_geo": "global",
  "iterations": [
   {
    "input_tokens": 2,
    "output_tokens": 447,
    "cache_read_input_tokens": 20592,
    "cache_creation_input_tokens": 472,
    "cache_creation": {
     "ephemeral_5m_input_tokens": 0,
     "ephemeral_1h_input_tokens": 472
    },
    "type": "message"
   }
  ],
  "speed": "standard",
  "fallback_credit": null
 },
 "modelUsage": {
  "claude-opus-5-5": {
   "inputTokens": 8,
   "outputTokens": 1367,
   "cacheReadInputTokens": 69651,
   "cacheCreationInputTokens": 8918,
   "webSearchRequests": 0,
   "costUSD": 0.1126462,
   "contextWindow": 1000000,
   "maxOutputTokens": 128000,
   "thinkingTokens": 120,
   "canonicalModel": "claude-opus-5-5",
   "provider": "firstParty",
   "costBasis": "list"
  }
 },
 "permission_denials": [],
 "terminal_reason": "completed",
 "fast_mode_state": "off",
 "fast_mode_disabled_reason": "sdk_opt_in_required",
 "subagent_stats": {
  "spawned": 0,
  "requested": {
   "background": 0,
   "foreground": 0,
   "unset": 0
  },
  "started_in_background": 0,
  "max_depth": 0,
  "spawned_by_subagents": 0,
  "completed": 0,
  "failed": 0,
  "killed": {
   "parent": 0,
   "user": 0,
   "system": 0
  },
  "refused": {
   "depth_limit": 0,
   "concurrency_limit": 0,
   "budget": 0
  },
  "by_type": {}
 },
 "safety_stops": 0,
 "is_error": false,
 "num_turns": 4,
 "subtype": "success",
 "api_error_status": null,
 "result": "I wrote `choice.json`. The cheapest flight that meets every constraint is **F006**: Coastal, LGA \u2192 SFO on Oct 5, business class, nonstop, leaving at 13:00, for $1,263.\n\nFour of the 49 flights in the file qualify:\n\n| Rank | ID | Route | Date | Cabin | Departs | Price |\n|---|---|---|---|---|---|---|\n| Best | F006 | LGA\u2192SFO | Oct 5 | business | 13:00 | $1,263 |\n| Runner-up | F025 | LGA\u2192SFO | Oct 6 | business | 21:00 | $1,395 |\n| Runner-up | F017 | JFK\u2192SFO | Oct 6 | business | 10:00 | $1,532 |\n| \u2014 | F003 | JFK\u2192SJC | Oct 5 | business | 16:30 | $1,687 |\n\nNone of the qualifying flights are first class, and all three picks fly into SFO rather than SJC.\n\nToday is October 14, 2026, so October 5 and 6 have already passed. That doesn't matter for a mock search, but if you meant future dates, give me the new dates and I'll run it again.",
 "ttft_ms": 1935,
 "type": "result",
 "duration_ms": 13956,
 "uuid": "[REDACTED]",
 "ttft_stream_ms": 776,
 "time_to_request_ms": 210,
 "first_content_frame_ms": 776,
 "queued_turn_count": 0,
 "result_index": 0
}

Get new runs and ranking changes weekly