Can Dify handle conflicting retrieved sources?
Dify flagged conflicting warranty terms rather than choosing one as fact. It retrieved documents stating 24 months and 36 months from delivery and included both periods in its answer.
1 of 1 test case passed
Every test case in this scenario has a result.
The test case
Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above is based on this test case.
Two documents give different values for the same factPassEvidence
“What is the warranty period on a new Tolvan A3?”
I’m not sure—some versions list 24 months and others 36 months from the delivery date. Could you share your purchase date or the warranty terms/version from your sales pack so I can confirm the correct period for your instrument?
Request and response log: the same exchange with the app settings appended
Copied from Captured file (text), line 20
Open the text file ↗| ✓ | Expected: Convey that the retrieved sources disagree, without presenting either warranty period as the definitive fact.Found: The conflicting documents ranked first and second, stating 36 and 24 months from delivery; the answer named both periods and chose neither. |
An uncertain answer alone would not establish that Dify recognized conflicting sources: both documents had to be retrieved. Dify retrieved both and acknowledged their disagreement without treating either warranty period as settled fact.
Suggested clarification. For the warranty question, Dify asked for the customer's purchase date or the warranty version in the sales pack to help resolve the disagreement.
Unrelated interval left out. The third-ranked sales-pack passage gave no warranty figure. The fourth-ranked product overview gave an unrelated 22-month recalibration interval, which did not appear in the answer.
Configuration and setup
How this tool was set up for the run and what the test needed in place. Each row is a fact from the run's records; a fact the records do not hold is left out, not guessed.
How this scenario is graded
How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.
Each test case gets one result per tool: Pass, Fail or Not gradable. A test case we haven't run yet shows Untested. There are no partial results.
- Pass: the tool did everything the test expected, and nothing it said contradicts the correct answer.
- Fail: at least one expected behaviour clearly didn't happen; the test case's result says which one and quotes the tool.
- Not gradable: our evidence couldn't settle the outcome (for example a record we needed is missing). It is never counted as a fail, and the test case's result says what's missing.
Where this sits in the benchmark
This page is one cell of a larger study: one tool, one scenario. Only this benchmark's frame appears here.
| Level | Name | Scope |
|---|---|---|
| Benchmark | Managed RAG → | 15 scenarios · 9 tools |
| Capability | Grounded Generation → | |
| Scenario | The retrieved sources disagree with each other → | |
| Tool | Dify → |
Global scenario definition → · Global capability definition → · Dify product page →
History of this result
What has happened to this result since it was first published. Runs and grades are never overwritten: a retest or a re-grade publishes a new result and keeps the earlier one readable.
Act on this result
Nothing filed here edits the run or the grade. A challenge opens a review, and a review can produce a new run or a re-grade — which becomes the current result and leaves this one in the history.
You run the same kind of test against your own setup and get the same behaviour.
Agree →Yours behaves differently. Tell us what you got, with a screenshot if you have one.
Disagree →Something here is wrong — a reference value, a transcription, a grade.
Report an issue →Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.
Vendor notice →The same record is available as structured data through the AI Demos MCP server, with the counts, the coverage and every per-test-case reason carried as fields.
Verify the proof files
These files support this result. Open a file to inspect the original evidence.
File fingerprints (SHA-256)
A fingerprint identifies the exact file used for this result.