Claude Sonnet 5.5 (effort default (no --effort flag) · thinking on) on Model ladder: fix 10 real merged bugs with tests hidden

[
 {
  "duration_api_ms": 29954,
  "stop_reason": "end_turn",
  "session_id": "[REDACTED]",
  "total_cost_usd": 0.09399410000000001,
  "usage": {
   "input_tokens": 16,
   "cache_creation_input_tokens": 12143,
   "cache_read_input_tokens": 159201,
   "output_tokens": 2947,
   "output_tokens_details": {
    "thinking_tokens": 459
   },
   "server_tool_use": {
    "web_search_requests": 0,
    "web_fetch_requests": 0
   },
   "service_tier": "standard",
   "cache_creation": {
    "ephemeral_1h_input_tokens": 12143,
    "ephemeral_5m_input_tokens": 0
   },
   "inference_geo": "global",
   "iterations": [
    {
     "input_tokens": 2,
     "output_tokens": 393,
     "cache_read_input_tokens": 24411,
     "cache_creation_input_tokens": 271,
     "cache_creation": {
      "ephemeral_5m_input_tokens": 0,
      "ephemeral_1h_input_tokens": 271
     },
     "type": "message"
    }
   ],
   "speed": "standard",
   "fallback_credit": null
  },
  "modelUsage": {
   "claude-sonnet-5-5": {
    "inputTokens": 16,
    "outputTokens": 2947,
    "cacheReadInputTokens": 159201,
    "cacheCreationInputTokens": 12143,
    "webSearchRequests": 0,
    "costUSD": 0.09399410000000001,
    "contextWindow": 1000000,
    "maxOutputTokens": 128000,
    "thinkingTokens": 459,
    "canonicalModel": "claude-sonnet-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": 11,
  "subtype": "success",
  "api_error_status": null,
  "result": "`intcomma(10**400 + 123)` and `intcomma(-(10**400 + 123))` now return the full digits, correctly grouped. The positive one ends in `,123`, and the negative one starts with `-10,000,`.\n\nThe cause was that `intcomma` called `float(value)` on every input to check it was finite. That raises `OverflowError` for ints too big for a float. The fix is in `src/humanize/number.py`: ints now skip the float conversion, and only floats get the finite check.\n\nStrings made of digits are treated as ints too. `intcomma(\"1\" + \"0\"*400)` used to be reported as infinity and now formats normally. Other string inputs behave as before: `\"1e3\"` and `\"abc\"` come back unchanged, and `\"inf\"` still gives `+Inf`.\n\nI added `test_intcomma_huge_int` to `tests/test_number.py`. The full suite gives 710 passed, 110 skipped and 15 errors. The same 15 errors appear without my change, all in `tests/test_benchmarks.py`; I didn't look into them.\n\nI only checked huge ints without `ndigits`. I didn't test `intcomma` on a huge int together with `ndigits`.",
  "ttft_ms": 2260,
  "type": "result",
  "duration_ms": 35874,
  "uuid": "[REDACTED]",
  "ttft_stream_ms": 1682,
  "time_to_request_ms": 293,
  "first_content_frame_ms": 1682,
  "queued_turn_count": 0,
  "result_index": 0
 }
]