Exa in Web Search for AI
Scenario-level performance from current published Results.
4 of 10 scenarios published
Exa passed every graded scenario it has a published Result for.
How Exa performed
Open a capability to explore its scenarios. Each row reports the test set in its published Result.
Freshness1 pass2 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| A new page has just gone live | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedExa's search service found the newly launched Ostrell oak sideboard page at rank 1, with its $1185.00 price, about 40 hours after it went live. Before launch it found nothing from the store, and two searches about 4.5 and 4.7 hours after launch returned no page from the store either. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| A page the tool already finds has changed | No published result | ||
| Content is removed from the source | No published result | ||
Web Retrieval3 pass1 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| A page on the web answers the question | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedExa answered the question by returning pages that state the Files API upload limit is 512 MB. The OpenAI Files API reference is at rank 1, and all ten returned results carry that figure; the other sizes mentioned on those pages are different limits. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| Search only these websites | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedExa met the site-only request. It returned docs.github.com results, including the Copilot page at rank 2, and no GitHub blog result; a rank-5 preview-policy page mentions other dates but not the Copilot launch year. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| Something more famous has the same name | 1 Pass0 Fail0 Not gradable 1 of 1 test case passedWhat happenedExa returned ten pages about Grafana Labs’ Loki and none about Marvel’s Loki. The returned documentation also identified the required compactor configuration block for enabling retention. | 1/1 assessed1/1 gradablePublished test set | View Result → |
| The request uses different wording than the source | No published result | ||
Answer Generation3 no Result yet
Capability in this benchmark →
| Scenario | Published outcomes | Test coverage | Result |
|---|---|---|---|
| The retrieved material doesn't support an answer | No published result | ||
| The retrieved material supports an answer | No published result | ||
| The retrieved sources disagree with each other | No published result | ||
Reading these Results
Published evidence and test coverage answer different questions.
Which scenarios have a Result?
A published Result is public evidence for this tool on one scenario. “No published result” does not say whether testing has taken place.
What does each Result cover?
Assessed includes Pass, Fail and Not gradable. Gradable includes Pass and Fail. Both use the pinned test count in that published Result.
Out of 10 scenarios
The summary counts each of the 10 scenarios once. A scenario fails when any of its published test cases failed. “No published Result yet” means nothing is public for that scenario; it does not say whether Exa was tested on it.