# Open-Ended Question Answering on SAGE Web Search Source: [Open-Ended Question Answering on SAGE Web Search]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:08:49.274Z Evaluation of model performance on the SAGE Web Search dataset for Open-Ended Question Answering, reporting weighted recall across specific subject domains (e.g., Computer Science, Healthcare) and overall average performance. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Open-Ended Question Answering | | Dataset (including any stated split) | SAGE Web Search | | Metric | Weighted Recall (Com. Sci.) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Weighted Recall (Com. Sci.)](): Higher is better; unit: PERCENT. - [Weighted Recall (Healthcare)](): Higher is better; unit: PERCENT. - [Weighted Recall (Humanities)](): Higher is better; unit: PERCENT. - [Weighted Recall (Nat. Sci.)](): Higher is better; unit: PERCENT. - [Average Performance](): Higher is better; unit: PERCENT. ## Recorded results | Method | Weighted Recall (Com. Sci.) | Result date | Report paper | | --- | --- | --- | --- | | GPT-5 | 35.1 | Not recorded | 698575dfc97a0b3e3b5d5eec | | gte-Qwen | 28.9 | Not recorded | 698575dfc97a0b3e3b5d5eec | | Gemini-2.5-flash | 28.2 | Not recorded | 698575dfc97a0b3e3b5d5eec | | GPT-5-mini | 27.4 | Not recorded | 698575dfc97a0b3e3b5d5eec | | GPT-5-nano | 25.9 | Not recorded | 698575dfc97a0b3e3b5d5eec | | gte-Qwen | 22.4 | Not recorded | 698575dfc97a0b3e3b5d5eec | | BM25 | 20.3 | Not recorded | 698575dfc97a0b3e3b5d5eec | | Gemini-2.5-pro | 20 | Not recorded | 698575dfc97a0b3e3b5d5eec | | BM25 | 18.2 | Not recorded | 698575dfc97a0b3e3b5d5eec | | DR Tulu | 18 | Not recorded | 698575dfc97a0b3e3b5d5eec | | ReasonIR | 16.1 | Not recorded | 698575dfc97a0b3e3b5d5eec | | ReasonIR | 11.3 | Not recorded | 698575dfc97a0b3e3b5d5eec | ### GPT-5 Weighted Recall (Com. Sci.): 35.1. Result ID: 6998a38c5698665ec4061219. [Reporting paper]() Recorded evaluation configuration: {"Avg. Searches":"13.69","Avg. Refs":"29.02"}. This record does not assert independent reproduction. ### gte-Qwen Weighted Recall (Com. Sci.): 28.9. Result ID: 6998a38c5698665ec4061225. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"k":"10","Avg. Searches":"4.54","Avg. Refs":"29.3"}. This record does not assert independent reproduction. ### Gemini-2.5-flash Weighted Recall (Com. Sci.): 28.2. Result ID: 6998a38c5698665ec406121d. [Reporting paper]() Recorded evaluation configuration: {"Avg. Searches":"6.50","Avg. Refs":"22.12"}. This record does not assert independent reproduction. ### GPT-5-mini Weighted Recall (Com. Sci.): 27.4. Result ID: 6998a38c5698665ec406121a. [Reporting paper]() Recorded evaluation configuration: {"Avg. Searches":"10.07","Avg. Refs":"30.70"}. This record does not assert independent reproduction. ### GPT-5-nano Weighted Recall (Com. Sci.): 25.9. Result ID: 6998a38c5698665ec406121b. [Reporting paper]() Recorded evaluation configuration: {"Avg. Searches":"8.59","Avg. Refs":"30.27"}. This record does not assert independent reproduction. ### gte-Qwen Weighted Recall (Com. Sci.): 22.4. Result ID: 6998a38c5698665ec4061228. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"k":"5","Avg. Searches":"4.79","Avg. Refs":"15.4"}. This record does not assert independent reproduction. ### BM25 Weighted Recall (Com. Sci.): 20.3. Result ID: 6998a38c5698665ec4061226. [Reporting paper]() Recorded evaluation configuration: {"k":"10","Avg. Searches":"4.17","Avg. Refs":"29.9"}. This record does not assert independent reproduction. ### Gemini-2.5-pro Weighted Recall (Com. Sci.): 20. Result ID: 6998a38c5698665ec406121e. [Reporting paper]() Recorded evaluation configuration: {"Avg. Searches":"10.77","Avg. Refs":"14.18"}. This record does not assert independent reproduction. ### BM25 Weighted Recall (Com. Sci.): 18.2. Result ID: 6998a38c5698665ec4061229. [Reporting paper]() Recorded evaluation configuration: {"k":"5","Avg. Searches":"4.72","Avg. Refs":"16.7"}. This record does not assert independent reproduction. ### DR Tulu Weighted Recall (Com. Sci.): 18. Result ID: 6998a38c5698665ec406121c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Avg. Searches":"4.25","Avg. Refs":"35.95"}. This record does not assert independent reproduction. ### ReasonIR Weighted Recall (Com. Sci.): 16.1. Result ID: 6998a38c5698665ec4061227. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"k":"10","Avg. Searches":"4.44","Avg. Refs":"26.8"}. This record does not assert independent reproduction. ### ReasonIR Weighted Recall (Com. Sci.): 11.3. Result ID: 6998a38c5698665ec406122a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"k":"5","Avg. Searches":"6.21","Avg. Refs":"15.8"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.