run_2026-09-14_nightly
Nightly · v3.4 · 2026-09-14
Claude Sonnet 5.5 · bge-m3 · hybrid (BM25 + dense, RRF) · support-prompt v7 on support-golden-v4 v4.6
Stratified slice of support-golden-v4 (419 items).
Pass rate
87.6%
Faithfulness
0.91
Answer correctness
0.88
Context precision
0.80
Context recall
0.84
Latency p95
3.8s
Failures by category
52 failed samples · largest share: unsupported
| Category | Share | vs reference |
|---|---|---|
| Unsupported claims | 26.9% | +2.4 pts |
| Incorrect factual answers | 19.2% | −1.2 pts |
| Missing context | 19.2% | +0.9 pts |
| Retrieval mismatch | 17.3% | −1.1 pts |
| Instruction-following failures | 11.5% | −0.7 pts |
| Other | 5.8% | −0.4 pts |
Run details
- Dataset
- support-golden-v4 v4.6 · 419 items
- Generator
- Claude Sonnet 5.5
- Retriever
- bge-m3 · hybrid (BM25 + dense, RRF)
- Prompt
- support-prompt v7
- Judge
- LLM judge · Claude Opus 5.5 · rubric v3
- Started
- Sep 14, 02:00 UTC
- Duration
- 44m 32s
- Triggered by
- Scheduler
- Tokens / request
- 3,420 in · 214 out
- Compared with
- Nightly · v3.4 · 2026-09-13
- nightly
- golden-slice
All metrics
Against Nightly · v3.4 · 2026-09-13
| Metric | Group | This run | Reference | Change |
|---|---|---|---|---|
| Pass rateShare of samples the judge marked as passing all rubric checks. | Outcome | 87.6% | 87.9% | −0.3 pts |
| FaithfulnessShare of answer claims supported by the retrieved context. | Generation | 0.91 | 0.91 | 0.00 |
| Answer correctnessAgreement with the gold answer on facts, not style. | Generation | 0.88 | 0.88 | 0.00 |
| Instruction adherenceRespect for tone, format, length, language and policy constraints. | Generation | 0.93 | 0.92 | +0.01 |
| Context precisionRank-weighted precision of relevant chunks passed to the generator. | Retrieval | 0.80 | 0.79 | +0.01 |
| Context recallShare of gold-answer statements supported by the retrieved context. | Retrieval | 0.84 | 0.85 | −0.01 |
| Recall@5Share of gold passages found in the top 5 retrieved chunks. | Retrieval | 0.89 | 0.88 | +0.02 |
| MRRMean reciprocal rank of the first relevant chunk. | Retrieval | 0.74 | 0.73 | +0.01 |
| NDCG@10Graded ranking quality of the top 10, normalized by the ideal order. | Retrieval | 0.81 | 0.80 | +0.01 |
| Latency p50Median end-to-end time to final token. | Operational | 2.0s | 2.0s | 0.0s |
| Latency p9595th percentile end-to-end time to final token. | Operational | 3.8s | 3.6s | +0.2s |
| Cost per requestModel, embedding and reranking cost per answered request (USD). | Operational | $0.0041 | $0.0043 | −0.0002 |
Nightly trend
Nightly runs on the support-golden-v4 slice, Jul 11 to Oct 8. This run is one point in the series.
Across datasets
support-assistant v3.4 on every dataset in the project. Rows marked “matrix” come from the comparison matrix rather than a dedicated run.
| Dataset | Items | Pass rate | Faith. | Ctx recall | Recall@5 | p95 | Source |
|---|---|---|---|---|---|---|---|
| support-golden-v4this run | 1,240 | 87.6% | 0.91 | 0.84 | 0.89 | 3.8s | This run |
| billing-edge-cases | 312 | 81.7% | 0.87 | 0.82 | 0.86 | 4.0s | Run · Oct 2 |
| roaming-and-travel | 188 | 80.9% | 0.88 | 0.79 | 0.83 | 3.9s | Run · Sep 28 |
| device-troubleshooting | 426 | 84.5% | 0.89 | 0.83 | 0.87 | 4.5s | Run · Sep 30 |
| account-security | 154 | 88.3% | 0.92 | 0.85 | 0.89 | 3.7s | Run · Sep 25 |
| locale-formatting | 96 | 80.2% | 0.86 | 0.80 | 0.84 | 4.1s | Run · Oct 4 |
| synthetic-hard-negatives | 540 | 76.5% | 0.86 | 0.73 | 0.77 | 4.0s | matrix |
Traced samples
17 samples with full traces (4 failing, 13 passing), stratified by domain. Open a row to inspect retrieval and the judge's reasoning.
How long does it take to port my number to Harbor? Number porting Passjudge 0.93 |
Does the Fall 2026 upgrade offer work with a family plan discount? Promotions Passjudge 0.92 |
What is the excess on a theft insurance claim? Returns & warranty Passjudge 0.90 |
If I upgrade my plan mid-month, how is my next bill calculated? Plans & upgrades Passjudge 0.96 |
How much data can I use in the EU on the Unlimited Plus plan? Roaming & travel Passjudge 0.97 |
Why was I charged twice for my September bill? Billing & payments Passjudge 0.94 |
Can I return a phone I bought online if I opened the box? Returns & warranty Passjudge 0.87 |
My bill says 41,20 €. Is that forty-one euros or four thousand? Billing & payments Passjudge 0.95 |
My eSIM QR code says it has already been used. What should I do? eSIM & activation Passjudge 0.87 |
Is SIM swap protection turned on by default? Account & security Passjudge 0.93 |