Jev Labs

Receipt Fields

Pick amounts already on the receipt. Recorded results below are from openjev, a separate model, not TypeSafe Jev.

One call: 35/50 receipts with all five fields correct on the primary seeded test cohort, compared with 27/50 for code rules and 32/50 for staged calls. This is openjev (Codiv free hosted) - NOT JEV, not a Jev result.

In a separate 100-receipt extension, one call scored 64/93 fully scorable receipts. Seven receipts have incomplete gold labels; this extension is not pooled with the primary comparison. The real-Jev free-route run is partial and has no completed receipts to compare. See the full methods and source tables.

Details · results, methods & full write-up
← All projects

Project 15

Receipt Fields

pick amounts already on the receipt

Details

Results

Generated from saved artifacts; intervals are Wilson 95% intervals. The opening and the following primary tables retain the original seeded cohort. RESULTS.md includes present-only scores, paired comparisons, development results and failure examples.

openjev (Codiv free hosted) - NOT JEV · full · v2_codiv_full_r1.json
ArmtotalsubtotaltaxcashchangeAll fields
Code rule42/50 (84.0%; 71.5%–91.7%)40/50 (80.0%; 67.0%–88.8%)48/50 (96.0%; 86.5%–98.9%)42/50 (84.0%; 71.5%–91.7%)45/50 (90.0%; 78.6%–95.7%)27/50 (54.0%; 40.4%–67.0%)
One call48/50 (96.0%; 86.5%–98.9%)47/50 (94.0%; 83.8%–97.9%)41/50 (82.0%; 69.2%–90.2%)49/50 (98.0%; 89.5%–99.6%)50/50 (100.0%; 92.9%–100.0%)35/50 (70.0%; 56.2%–80.9%)
One call per field47/50 (94.0%; 83.8%–97.9%)44/50 (88.0%; 76.2%–94.4%)44/50 (88.0%; 76.2%–94.4%)48/50 (96.0%; 86.5%–98.9%)46/50 (92.0%; 81.2%–96.8%)32/50 (64.0%; 50.1%–75.9%)
Always absent3/50 (6.0%; 2.1%–16.2%)18/50 (36.0%; 24.1%–49.9%)31/50 (62.0%; 48.2%–74.1%)17/50 (34.0%; 22.4%–47.8%)23/50 (46.0%; 33.0%–59.6%)0/50 (0.0%; 0.0%–7.1%)
REAL JEV (OpenCode Zen free, jev-1.13-free) · zen · v2_zen_zen_r1.json
ArmtotalsubtotaltaxcashchangeAll fields
Code rulenot measurednot measurednot measurednot measurednot measurednot measured
One callnot measurednot measurednot measurednot measurednot measurednot measured
One call per fieldnot measurednot measurednot measurednot measurednot measurednot measured
Always absentnot measurednot measurednot measurednot measurednot measurednot measured
Secondary: complete official test split

This separately registered openjev (Codiv free hosted) - NOT JEV extension covers the official test split and overlaps the original seeded cohort. It is reported separately, never pooled with the primary comparison. Prompts, candidates and code rules remain frozen. Unknown gold is excluded only from the affected field scores and from fully scorable receipt scores; every receipt remains in the execution plan.

openjev (Codiv free hosted) - NOT JEV · full100 · v2_codiv_full100_r1.json
ArmtotalsubtotaltaxcashchangeAll fields
Code rule83/100 (83.0%; 74.5%–89.1%)81/99 (81.8%; 73.1%–88.2%)92/97 (94.8%; 88.5%–97.8%)82/97 (84.5%; 76.0%–90.4%)87/100 (87.0%; 79.0%–92.2%)49/93 (52.7%; 42.6%–62.5%)
One call94/100 (94.0%; 87.5%–97.2%)89/99 (89.9%; 82.4%–94.4%)83/97 (85.6%; 77.2%–91.2%)94/97 (96.9%; 91.3%–98.9%)97/100 (97.0%; 91.5%–99.0%)64/93 (68.8%; 58.8%–77.3%)
One call per field95/100 (95.0%; 88.8%–97.8%)78/99 (78.8%; 69.7%–85.7%)91/97 (93.8%; 87.2%–97.1%)92/97 (94.8%; 88.5%–97.8%)94/100 (94.0%; 87.5%–97.2%)58/93 (62.4%; 52.2%–71.5%)
Always absent5/100 (5.0%; 2.2%–11.2%)35/99 (35.4%; 26.6%–45.2%)57/97 (58.8%; 48.8%–68.0%)33/97 (34.0%; 25.4%–43.9%)44/100 (44.0%; 34.7%–53.8%)1/93 (1.1%; 0.2%–5.8%)

Processed receipts saved: 100. A saved row can still have incomplete model answers.

Each field score above uses only answered, scorable labels. Fully scorable receipts have known gold for every field. The conservative bound awards no credit to receipts with unknown gold and includes every fully predicted receipt.

ArmFully predicted receiptsFully scorable receiptsConservative all-receipt bound
Code rule1009349/100 (49.0%; 39.4%–58.7%)
One call1009364/100 (64.0%; 54.2%–72.7%)
One call per field1009358/100 (58.0%; 48.2%–67.2%)
Always absent100931/100 (1.0%; 0.2%–5.4%)

Annotation exclusions (unknown gold, distinct from absence):

Every preregistered annotation issue from the frozen fixture manifest is listed, including receipts not yet processed. See FULL100_MANIFEST.json and the amendment in the registered protocol.

ReceiptFieldRaw annotationReasonProcessed row saved
cord_test_013subtotal["46.636", "46.636"]unrecognized amountyes
cord_test_058cash["100,000", "100,000"]unrecognized amountyes
cord_test_060tax"-"unrecognized amountyes
cord_test_065tax"-"unrecognized amountyes
cord_test_076cash"*110000"unrecognized amountyes
cord_test_095cash["62,000", "201"]unrecognized amountyes
cord_test_099tax["6,818", "0"]unrecognized amountyes

Status: complete; stop: none; last-process wall: 602.738 s. Successful replayed calls are counted once.

Requests and cost
Route / artifactArmRequests ok / attemptedInput / output tokensHTTP p50 / p95 msCompleted receiptsFree-route costEstimated list-price cost / 1,000 receipts
openjev (Codiv free hosted) - NOT JEV / v2_codiv_dev_r1.jsonOne call20 / 2033191 / 0347 / 68920$0$0.0697
openjev (Codiv free hosted) - NOT JEV / v2_codiv_dev_r1.jsonOne call per field100 / 10063815 / 0287 / 52120$0$0.1340
openjev (Codiv free hosted) - NOT JEV / v2_codiv_full100_r1.jsonOne call100 / 100191075 / 0567 / 880100$0$0.0803
openjev (Codiv free hosted) - NOT JEV / v2_codiv_full100_r1.jsonOne call per field500 / 500366375 / 0400 / 825100$0$0.1539
openjev (Codiv free hosted) - NOT JEV / v2_codiv_full_r1.jsonOne call50 / 5094664 / 0332 / 52750$0$0.0795
openjev (Codiv free hosted) - NOT JEV / v2_codiv_full_r1.jsonOne call per field250 / 250181220 / 0287 / 39550$0$0.1522
REAL JEV (OpenCode Zen free, jev-1.13-free) / v2_zen_zen_r1.jsonOne call0 / 20 / 0n/a0$0n/a
REAL JEV (OpenCode Zen free, jev-1.13-free) / v2_zen_zen_r1.jsonOne call per field0 / 00 / 0n/a0$0n/a

The estimate uses the shared client's existing TypeSafe input-token list price, output free. It is not a bill for either free route. Failed-attempt usage is unknown. HTTP latency excludes lock waits and cooldown. For partial runs, successful field calls on incomplete receipts remain in tokens; the cost denominator is completed predictions, including receipts with unscorable gold. Unreported failed-request usage cannot be reconstructed.

Receipt stop history

Each row is a distinct saved stop. Archived copies of the current stop appear once. Counts are cumulative at that stop and are not added to request totals or counted as repeats. HTTP status and provider wording appear only when recorded; a failure class does not supply them. Attempt starts are runner timestamps, before any client pacing or lock waits.

Route / artifactAttempt started (UTC)Stopped at (UTC)Completed predictions: one call / stagedRequests ok / attemptedSaved HTTP statusFailure classProvider wording
REAL JEV (OpenCode Zen free, jev-1.13-free) / interrupted/v2_zen_zen_r1_1.json2026-09-24 22:26:01 UTC2026-09-24 22:26:17 UTC0 / 00 / 1429rate_limitretained provider wording: “Rate limit exceeded. Please try again later.”
REAL JEV (OpenCode Zen free, jev-1.13-free) / v2_zen_zen_r1.json2026-09-24 23:28:55 UTC2026-09-24 23:29:11 UTC0 / 00 / 2429rate_limitretained provider wording: “Rate limit exceeded. Please try again later.”
Run
./run.sh --plan
./run.sh --offline --repeat-start "$REPEAT"
./run.sh --mock --repeat-start "$REPEAT"
Limits and provenance

ATTRIBUTION.md documents CORD and CC-BY-4.0. Only annotated valid-line text is used. No image is downloaded and OCR errors are outside this experiment. Missing labels count as absent; currency normalization and candidate coverage can limit accuracy. Rules were inspected only on dev. The primary test cohort is scored once per route; the separately registered secondary pass makes fresh calls on the overlapping official test split. Unknown secondary annotations are unscorable, never treated as absent. The fixed Zen staged subset has fewer receipts, so use the matched results when comparing arms.

The primary comparison uses a seeded sample of the official test split. The shared cohort keeps the route comparison within the registered Zen call budget; openjev uses the same cohort. The extension is a separate secondary measurement with annotation exclusions registered before its model calls. See the sampling clarification and amendment in the registered protocol; the original cohort and its prompt and scoring rules are preserved.

Raw result files
CODE ONLY - NO MODELresults/v2_code_full100_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "full100",
 "repeat": 1,
 "limit": null,
 "campaign": null,
 "timestamp": "2026-09-24T20:54:57Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/FULL100_MANIFEST.json": "612fe780d245f9075349e7048a4ca5416c2ea66cc76427de11a5fef3f67b4e93",
  "samples/receipt-fields/fixtures/full100.jsonl": "01ad0f2583499782d5b589d8323e9665e5c26432e262198396dfc235a80dd301"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "5d632a05dec5069caaa2be83198f7c5a193bd5e85dd876b1ef2d118f4337809d",
  "samples/receipt-fields/../_shared/community.py": "4d727a7ef8b691d127a40dbba6ed8350a024dbc8a22364b5d877e8ea947295fc",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/build_full_test.py": "2ce9ff55be73e2ace0882d381e428333f6108843c5a8a57e92487f502081d619",
  "samples/receipt-fields/receipt_fields.py": "d2068eeafdb6c34e6e8dfdda48e6c4bb98635635ae666e920f558bc14c1afb1c",
  "samples/receipt-fields/report.py": "2c386338050b8e0ad4061e8ca538dce1568494122c067e01a4d6b9196bd49e6f"
 },
 "label": "code only (no model)",
 "mode": "code",
 "provider": null,
 "call_cap": null,
 "rows": [
  {
   "id": "cord_test_000",
   "repeat": 1,
   "gold": {
    "total": "60000",
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "code": {
    "total": "-60000",
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 3,
   "candidate_gold_hits": 3,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_001",
   "repeat": 1,
   "gold": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "code": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 2,
   "candidate_gold_hits": 2,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_002",
   "repeat": 1,
   "gold": {
    "total": "28000",
    "subtotal": "28000",
    "tax": null,
    "cash": "100000",
    "change": "72000"
   },
   "code": {
    "total": "28000",
    "subtotal": "28000",
    "tax": null,
    "cash": "100000",
    "change": "72000"
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 4,
   "candidate_gold_hits": 4,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_003",
   "repeat": 1,
   "gold": {
    "total": "11000",
    "subtotal": "11000",
    "tax": null,
    "cash": "11000",
    "change": null
   },
   "code": {
    "total": "11000",
    "subtotal": "11000",
    "tax": null,
    "cash": "11000",
    "change": null
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 3,
   "candidate_gold_hits": 3,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_004",
   "repeat": 1,
   "gold": {
    "total": "174600",

… truncated, 2994 of 55747 bytes shown. Complete safe artifact

CODE ONLY - NO MODELresults/v2_code_full_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "full",
 "repeat": 1,
 "limit": null,
 "timestamp": "2026-09-24T20:08:22Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/cases.jsonl": "bbccf4ef74097f307744b4ad1c27d7ab50859597e5aeaf9cf7fc7ce94ae0ae8a"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "5d632a05dec5069caaa2be83198f7c5a193bd5e85dd876b1ef2d118f4337809d",
  "samples/receipt-fields/../_shared/community.py": "4752e0ac6bed9c3cce38c0ba9713d4c9cbc22c966774663950916e7fc559bd81",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/receipt_fields.py": "1ba39e82e5f8309fff44a4af9d0318e49e6554561b6d06ac56e33ac05db38ba5",
  "samples/receipt-fields/report.py": "bc2d406b358b716ea8c31d196505fc7252c7a1278050194722921fc7aee52728"
 },
 "label": "code only (no model)",
 "mode": "code",
 "provider": null,
 "call_cap": null,
 "rows": [
  {
   "id": "cord_test_078",
   "repeat": 1,
   "gold": {
    "total": "30000",
    "subtotal": null,
    "tax": null,
    "cash": "30000",
    "change": "0"
   },
   "code": {
    "total": "30000",
    "subtotal": null,
    "tax": null,
    "cash": "30000",
    "change": null
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 3,
   "candidate_gold_hits": 3
  },
  {
   "id": "cord_test_089",
   "repeat": 1,
   "gold": {
    "total": "40000",
    "subtotal": "40000",
    "tax": null,
    "cash": "50000",
    "change": "10000"
   },
   "code": {
    "total": "40000",
    "subtotal": null,
    "tax": null,
    "cash": "10000",
    "change": "10000"
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 4,
   "candidate_gold_hits": 4
  },
  {
   "id": "cord_test_096",
   "repeat": 1,
   "gold": {
    "total": "41000",
    "subtotal": "41000",
    "tax": null,
    "cash": "50000",
    "change": "9000"
   },
   "code": {
    "total": null,
    "subtotal": null,
    "tax": null,
    "cash": "9000",
    "change": "9000"
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 4,
   "candidate_gold_hits": 4
  },
  {
   "id": "cord_test_083",
   "repeat": 1,
   "gold": {
    "total": "23000",
    "subtotal": "20909",
    "tax": "2091",
    "cash": "25000",
    "change": "2000"
   },
   "code": {
    "total": "23000",
    "subtotal": "20909",
    "tax": "2091",
    "cash": null,
    "change": "2000"
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 5,
   "candidate_gold_hits": 5
  },
  {
   "id": "cord_test_067",
   "repeat": 1,
   "gold": {
    "total": "9500",
    "subtotal": null,
    "tax": null,
    "cash": "9500",
    "change": null
   },
   "code": {
    "total": "9500",
    "subtotal": null,
    "tax": null,
    "cash": null,
    "change": null
   },
   "one_call": null,
   "staged": null,
   "present_gold_fields": 2,
   "candidate_gold_hits": 2
  },
  {
   "id": "cord_test_031",
   "repeat": 1,
   "gold": {
    "total": "113886",

… truncated, 2995 of 30872 bytes shown. Complete safe artifact

openjev (Codiv free hosted) - NOT JEVresults/v2_codiv_dev_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "dev",
 "repeat": 1,
 "limit": null,
 "timestamp": "2026-09-24T20:17:48Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/cases.jsonl": "bbccf4ef74097f307744b4ad1c27d7ab50859597e5aeaf9cf7fc7ce94ae0ae8a"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "5d632a05dec5069caaa2be83198f7c5a193bd5e85dd876b1ef2d118f4337809d",
  "samples/receipt-fields/../_shared/community.py": "4752e0ac6bed9c3cce38c0ba9713d4c9cbc22c966774663950916e7fc559bd81",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/receipt_fields.py": "1ba39e82e5f8309fff44a4af9d0318e49e6554561b6d06ac56e33ac05db38ba5",
  "samples/receipt-fields/report.py": "75c82fa9684f0a9d3eafe16e6c3ac082cdab2e24b97cd1fe4bfe1dbd089e06e3"
 },
 "label": "openjev (Codiv free hosted) - NOT JEV",
 "mode": "live",
 "provider": "codiv",
 "call_cap": 126,
 "rows": [
  {
   "id": "cord_validation_000",
   "repeat": 1,
   "gold": {
    "total": "45500",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "4500"
   },
   "code": {
    "total": "45500",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "4500"
   },
   "one_call": {
    "total": "45500",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "4500"
   },
   "staged": {
    "total": "45500",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "4500"
   },
   "present_gold_fields": 3,
   "candidate_gold_hits": 3
  },
  {
   "id": "cord_validation_001",
   "repeat": 1,
   "gold": {
    "total": "23000",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "27000"
   },
   "code": {
    "total": "23000",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "27000"
   },
   "one_call": {
    "total": "23000",
    "subtotal": null,
    "tax": null,
    "cash": "50000",
    "change": "27000"
   },
   "staged": {
    "total": "23000",
    "subtotal": "23000",
    "tax": null,
    "cash": "50000",
    "change": "27000"
   },
   "present_gold_fields": 3,
   "candidate_gold_hits": 3
  },
  {
   "id": "cord_validation_002",
   "repeat": 1,
   "gold": {
    "total": "20000",
    "subtotal": "18181",
    "tax": "1818",
    "cash": "100000",
    "change": "80000"
   },
   "code": {
    "total": "20000",
    "subtotal": "18181",
    "tax": "10",
    "cash": "100000",
    "change": "80000"
   },
   "one_call": {
    "total": "20000",
    "subtotal": "18181",
    "tax": null,
    "cash": "100000",
    "change": "80000"
   },
   "staged": {
    "total": "20000",
    "subtotal": "18181",
    "tax": "1818",
    "cash": "100000",
    "change": "80000"
   },
   "present_gold_fields": 5,
   "candidate_gold_hits": 5
  },
  {
   "id": "cord_validation_003",
   "repeat": 1,
   "gold": {
    "total": "28000",
    "subtotal": "28000",
    "tax": "0",
    "cash": "28000",
    "change": "0"
   },

… truncated, 2998 of 179559 bytes shown. Complete safe artifact

openjev (Codiv free hosted) - NOT JEVresults/v2_codiv_full100_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "full100",
 "repeat": 1,
 "limit": null,
 "campaign": "m3-receipt-codiv-full100",
 "timestamp": "2026-09-24T20:57:37Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/FULL100_MANIFEST.json": "612fe780d245f9075349e7048a4ca5416c2ea66cc76427de11a5fef3f67b4e93",
  "samples/receipt-fields/fixtures/full100.jsonl": "01ad0f2583499782d5b589d8323e9665e5c26432e262198396dfc235a80dd301"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "5d632a05dec5069caaa2be83198f7c5a193bd5e85dd876b1ef2d118f4337809d",
  "samples/receipt-fields/../_shared/community.py": "4d727a7ef8b691d127a40dbba6ed8350a024dbc8a22364b5d877e8ea947295fc",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/build_full_test.py": "2ce9ff55be73e2ace0882d381e428333f6108843c5a8a57e92487f502081d619",
  "samples/receipt-fields/receipt_fields.py": "d2068eeafdb6c34e6e8dfdda48e6c4bb98635635ae666e920f558bc14c1afb1c",
  "samples/receipt-fields/report.py": "2c386338050b8e0ad4061e8ca538dce1568494122c067e01a4d6b9196bd49e6f"
 },
 "label": "openjev (Codiv free hosted) - NOT JEV",
 "mode": "live",
 "provider": "codiv",
 "call_cap": 630,
 "rows": [
  {
   "id": "cord_test_000",
   "repeat": 1,
   "gold": {
    "total": "60000",
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "code": {
    "total": "-60000",
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "one_call": {
    "total": null,
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "staged": {
    "total": "60000",
    "subtotal": "60000",
    "tax": "5455",
    "cash": null,
    "change": null
   },
   "present_gold_fields": 3,
   "candidate_gold_hits": 3,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_001",
   "repeat": 1,
   "gold": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "code": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "one_call": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "staged": {
    "total": "91000",
    "subtotal": null,
    "tax": null,
    "cash": "91000",
    "change": null
   },
   "present_gold_fields": 2,
   "candidate_gold_hits": 2,
   "unscorable_fields": {}
  },
  {
   "id": "cord_test_002",
   "repeat": 1,
   "gold": {
    "total": "28000",
    "subtotal": "28000",
    "tax": null,
    "cash": "100000",
    "change": "72000"
   },
   "code": {
    "total": "28000",
    "subtotal": "28000",
    "tax": null,
    "cash": "100000",
    "change": "72000"
   },
   "one_call": {
    "total": "28000",
    "subtotal": "28000",
    "tax": null,
    "cash": "100000",
    "change": "72000"
   },
   "staged": {
    "total": "28000",
    "subtotal":

… truncated, 2994 of 876857 bytes shown. Complete safe artifact

openjev (Codiv free hosted) - NOT JEVresults/v2_codiv_full_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "full",
 "repeat": 1,
 "limit": null,
 "timestamp": "2026-09-24T20:19:47Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/cases.jsonl": "bbccf4ef74097f307744b4ad1c27d7ab50859597e5aeaf9cf7fc7ce94ae0ae8a"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "5d632a05dec5069caaa2be83198f7c5a193bd5e85dd876b1ef2d118f4337809d",
  "samples/receipt-fields/../_shared/community.py": "4752e0ac6bed9c3cce38c0ba9713d4c9cbc22c966774663950916e7fc559bd81",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/receipt_fields.py": "1ba39e82e5f8309fff44a4af9d0318e49e6554561b6d06ac56e33ac05db38ba5",
  "samples/receipt-fields/report.py": "75c82fa9684f0a9d3eafe16e6c3ac082cdab2e24b97cd1fe4bfe1dbd089e06e3"
 },
 "label": "openjev (Codiv free hosted) - NOT JEV",
 "mode": "live",
 "provider": "codiv",
 "call_cap": 315,
 "rows": [
  {
   "id": "cord_test_078",
   "repeat": 1,
   "gold": {
    "total": "30000",
    "subtotal": null,
    "tax": null,
    "cash": "30000",
    "change": "0"
   },
   "code": {
    "total": "30000",
    "subtotal": null,
    "tax": null,
    "cash": "30000",
    "change": null
   },
   "one_call": {
    "total": "30000",
    "subtotal": "25000",
    "tax": null,
    "cash": "30000",
    "change": "0"
   },
   "staged": {
    "total": "30000",
    "subtotal": null,
    "tax": null,
    "cash": "30000",
    "change": "0"
   },
   "present_gold_fields": 3,
   "candidate_gold_hits": 3
  },
  {
   "id": "cord_test_089",
   "repeat": 1,
   "gold": {
    "total": "40000",
    "subtotal": "40000",
    "tax": null,
    "cash": "50000",
    "change": "10000"
   },
   "code": {
    "total": "40000",
    "subtotal": null,
    "tax": null,
    "cash": "10000",
    "change": "10000"
   },
   "one_call": {
    "total": "40000",
    "subtotal": "40000",
    "tax": null,
    "cash": "50000",
    "change": "10000"
   },
   "staged": {
    "total": "40000",
    "subtotal": "40000",
    "tax": null,
    "cash": "50000",
    "change": "10000"
   },
   "present_gold_fields": 4,
   "candidate_gold_hits": 4
  },
  {
   "id": "cord_test_096",
   "repeat": 1,
   "gold": {
    "total": "41000",
    "subtotal": "41000",
    "tax": null,
    "cash": "50000",
    "change": "9000"
   },
   "code": {
    "total": null,
    "subtotal": null,
    "tax": null,
    "cash": "9000",
    "change": "9000"
   },
   "one_call": {
    "total": "41000",
    "subtotal": "41000",
    "tax": null,
    "cash": "50000",
    "change": "9000"
   },
   "staged": {
    "total": "41000",
    "subtotal": "41000",
    "tax": null,
    "cash": "50000",
    "change": "9000"
   },
   "present_gold_fields": 4,
   "candidate_gold_hits": 4
  },
  {
   "id": "cord_test_083",
   "repeat": 1,
   "gold": {
    "total": "23000",
    "subtotal": "20909",
    "tax": "2091",
    "cash": "25000",
    "change": "2000"
   },
   "code": {
    "total": "23000",

… truncated, 3000 of 441795 bytes shown. Complete safe artifact

REAL JEV (OpenCode Zen free, jev-1.13-free)results/v2_zen_zen_r1.json
{
 "schema_version": 2,
 "project": "receipt-fields",
 "subset": "zen",
 "repeat": 1,
 "limit": null,
 "campaign": "m3-receipt-zen",
 "timestamp": "2026-09-24T23:28:55Z",
 "input_hashes": {
  "samples/receipt-fields/fixtures/cases.jsonl": "bbccf4ef74097f307744b4ad1c27d7ab50859597e5aeaf9cf7fc7ce94ae0ae8a"
 },
 "source_hashes": {
  "samples/_shared/jev.py": "a20ab00f05c9fc838194f68b33f88b64bca3a0dd7e1ddf78d4c347bf9cdf3322",
  "samples/receipt-fields/../_shared/community.py": "79d28fe055b35dada9eeb70122da25ae3487e901c0d0992ba1270207ad553968",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/build_full_test.py": "2ce9ff55be73e2ace0882d381e428333f6108843c5a8a57e92487f502081d619",
  "samples/receipt-fields/receipt_fields.py": "d2068eeafdb6c34e6e8dfdda48e6c4bb98635635ae666e920f558bc14c1afb1c",
  "samples/receipt-fields/report.py": "e36834f47dafe551b706e526a343177b00549f74b1687d4e1c9c6efa202c194d"
 },
 "label": "REAL JEV (OpenCode Zen free, jev-1.13-free)",
 "mode": "live",
 "provider": "zen",
 "call_cap": 237,
 "resumed_after": "rate_limit",
 "original_source_hashes": {
  "samples/_shared/jev.py": "a20ab00f05c9fc838194f68b33f88b64bca3a0dd7e1ddf78d4c347bf9cdf3322",
  "samples/receipt-fields/../_shared/community.py": "79d28fe055b35dada9eeb70122da25ae3487e901c0d0992ba1270207ad553968",
  "samples/receipt-fields/build_fixtures.py": "98b2622a0ae1d209a69bf9c1117c28dbd2759c4a318784057fa70649d39a824b",
  "samples/receipt-fields/build_full_test.py": "2ce9ff55be73e2ace0882d381e428333f6108843c5a8a57e92487f502081d619",
  "samples/receipt-fields/receipt_fields.py": "d2068eeafdb6c34e6e8dfdda48e6c4bb98635635ae666e920f558bc14c1afb1c",
  "samples/receipt-fields/report.py": "e36834f47dafe551b706e526a343177b00549f74b1687d4e1c9c6efa202c194d"
 },
 "status": "partial",
 "rows": [],
 "calls": [
  {
   "case_id": "cord_test_078",
   "phase": "one_call",
   "state_hash": "16c88a5f94f3f6345fd9a3746060910dee63da8ac185706f89b97ab2f20a3688",
   "questions_hash": "ad4c6e4dbc8a4bb162867ba7353711b255a1f85ac818c5f08056f8d78122655a",
   "question_ids": [
    "total",
    "subtotal",
    "tax",
    "cash",
    "change"
   ],
   "status": "failed",
   "reason": "rate_limit",
   "provider_message": "Rate limit exceeded. Please try again later.",
   "http_status": 429,
   "attempted_at": 1790288761.7979457,
   "elapsed_s": 15.395
  },
  {
   "case_id": "cord_test_078",
   "phase": "one_call",
   "state_hash": "16c88a5f94f3f6345fd9a3746060910dee63da8ac185706f89b97ab2f20a3688",
   "questions_hash": "ad4c6e4dbc8a4bb162867ba7353711b255a1f85ac818c5f08056f8d78122655a",
   "question_ids": [
    "total",
    "subtotal",
    "tax",
    "cash",
    "change"
   ],
   "status": "failed",
   "reason": "rate_limit",
   "provider_message": "Rate limit exceeded. Please try again later.",
   "http_status": 429,
   "attempted_at": 1790292535.661722,
   "elapsed_s": 15.409
  }
 ],
 "metrics": {
  "requests": 2,

… truncated, 2993 of 3318 bytes shown. Complete safe artifact