In the current artificial intelligence-driven innovation era, the pace of knowledge growth is accelerating, and is hard to keep up with. While generative models are increasingly used to synthesize content, they often lack in information grounding. To address these peculiarities of our time, we propose Wyvern, a multi-agent framework for the automated generation of grounded, multimodal technical reports. Wyvern allows for the generation of multimodal outputs, integrating images, tables, and text with supporting references in a unified report. Additionally, a particular focus is placed on the grounding of the content, with the implementation of a claims auto-revision stage. We conduct a human evaluation study to assess the quality of our proposed framework. The results show that the figures' informativeness is perceived as superior to that of a recent baseline in 87% of cases. Furthermore, Wyvern's reports are rated as more useful than those produced by three alternative methods in 63% to 100% of instances. We also carry out automatic evaluations showing that Wyvern gains up to 2.3× in citation recall and 1.6× in citation precision with respect to the baselines.
Figures & tables
Figure 1: Overview of the main functionalities of Wyvern
Method
Images insertion
Claims verification
STORM Shao et al. (2024)
✗
✗
WebThinker Li et al. (2025)
✗
✗
WikiAutoGen Yang et al. (2025)
✓
✗
Wyvern (ours)
✓
✓
Table 1: Overview of expository text generators. Wyvern integrates multimodality and explicit claims verification and revision into a unified framework.
Figure 2: Overview of the key components of Wyvern, our proposed framework. The search module (left) retrieves relevant documents from the web. The report generation module (middle) handles of the writing of the report’s text and of the figures insertion. The grounding module (right) verifies if the claims are supported by the cited references, and possibly revises them.
Figure 3: Structure of the figures selection and insertion routine. The pipeline comprises the generation of the descriptions of all the images extracted from the sources, the selection of the most relevant figures, and their integration in the report.
Figure 4: Overview of the grounding verification pipeline. The grounding of the claims is evaluated with their associated references, and then possibly revised.
Figure 5: Relative grading of Wyvern with respect to selected baselines in the human evaluation study. Each cell reports the preference percentage for Wyvern over the baseline on a rubric. STORM and WebThinker cannot be compared on the Figures criteria, as they produce text-only reports.
Figure 6: Results of the absolute grading of Wyvern’s reports from the human evaluation study
Figure 7: Citation quality metrics computed on all the reports generated by the various methods (27 documents in total for ours, 9 for STORM and WikiAutoGen). The error bar shows the standard deviation.
Metric
Before (%)
After (%)
Gain
Cit. recall
60.92 ± 6.10
73.79 ± 5.44
1.21 ×
Cit. recall w/o uncited claims
84.34 ± 4.24
94.64 ± 1.55
1.12 ×
Cit. precision
67.15 ± 6.17
75.45 ± 7.54
1.12 ×
Table 2: Impact of the claims revision module on the citations quality metrics
Figure 8: Comparison of human ( y -axis) and automatic ( x -axis) evaluation results on the four rubrics with two different evaluator models. The diagonal line represents a 1:1 correspondence between human and automatic evaluation.
Appendix figures & tables6 assets
Supplementary material from the paper’s appendix.
Appendix
Preference rate for Wyvern
Question
vs STORM
vs WebThinker
vs WikiAutoGen
1
Which report has the most logical structure?
100.00%
37.50%
75.00%
2
Which report better provides relevant information with respect to the topic?
100.00%
62.50%
75.00%
3
Which report presents the best amount of technical details about the topic?
100.00%
75.00%
75.00%
4
Which report better presents both benefits/advantages and challenges/limitations related to the topic?
100.00%
50.00%
87.50%
5
Which report has a broader coverage of multiple aspects of the topic?
100.00%
100.00%
87.50%
Appendix
Table 3: Results of the relative assessment section of the human evaluation study. Each percentage value represents the preference rate in the pairwise comparison between Wyvern and one of the baselines on a specific rubric. Comparisons cannot be made on criteria concerning figures when considering STORM and WebThinker as the other method, as explained in the paper.
Question
Score
1
The structure of the report is logical.
3.87 ± 0.61
2
The report provides relevant information with respect to the topic.
4.26 ± 0.53
3
The report presents the right amount of technical details about the topic.
3.87 ± 0.85
4
The report presents both benefits/advantages and challenges/limitations related to the topic.
3.91 ± 0.78
5
The report has a broad coverage of multiple aspects of the topic.
4.35 ± 0.56
6
The explanations of the report are:
3.91 ± 0.88
Appendix
Table 4: Results of the absolute grading of the human evaluation study, reported with the mean and standard deviation of the given marks over the 23 questionnaires that were completed, out of the 27. Participants could assign a score in the [1,5]-range. For rubric (6), the interval extremes were associated with the adjectives “confusing" and “clear"; for rubric (7) with “biased" and “neutral"; and for rubric (14) with “not satisfying" and “very satisfying".
Rubrics
vs STORM
vs WebThinker
vs WikiAutoGen
Interest level
50.00%
50.00%
50.00%
Coherence & Organization
50.00%
50.00%
50.00%
Relevance & Focus
50.00%
50.00%
50.00%
Broad Coverage
50.00%
50.00%
50.00%
Appendix
Table 5: Results of the automatic relative assessment, performed using Prometheus 2-7B as evaluator model
Score
Rubrics
Evaluator model
STORM
WebThinker
WikiAutoGen
Wyvern
Prometheus-13B
4.00 ± 0.00
4.56 ± 0.50
4.00 ± 0.00
4.07 ± 0.47
Prometheus 2-7B
4.22 ± 0.92
4.44 ± 0.83
4.67 ± 0.67
4.85 ± 0.45
DeepSeek-V3
3.22 ± 0.63
4.44 ± 0.83
3.44 ± 0.68
4.85 ± 0.52
Interest Level
Qwen3-32B
3.33 ± 0.47
4.33 ± 0.67
3.67 ± 0.67
3.78 ± 0.50
Prometheus-13B
4.78 ± 0.42
4.44 ± 0.50
4.33 ± 0.47
3.96 ± 0.58
Appendix
Table 6: Results of the automatic absolute grading, performed with different evaluator models, namely Prometheus-13B, Prometheus 2-7B, DeepSeek-V3, and Qwen3-32B. The results are reported in terms of the mean and standard deviation of the scores over all the reports generated by a given method. Underlined, the method achieving the highest average score with a specific evaluator on a given criterion.
Figure 9: Input and completion tokens used by each module of Wyvern.
Figure 10: Input and completion tokens used by each agent of Wyvern.