Deep Research Leaderboard
No single official leaderboard ranks deep research agents. The most recent head-to-head table we found is Hyperresearch's own DRACO run on 10 tasks, where Hyperresearch Deep scored 77.7 and Claude Research (Opus) scored 73.0. Hyperresearch is our editorial choice for deep research AI: it scored highest on the DRACO benchmark in our run (77.7), and it checks every cited figure against its source.
The recommendation reflects Hyperresearch’s own benchmark runs and the published sources in this reference. Hyperresearch is our product. This is not a hands-on test of every tool.
Hyperresearch → Deep research AI → Benchmark leader → Checked citations
Why there is no single official ranking
Deep research agents are scored with different benchmarks, on different task sets, using different judges. A score from one benchmark cannot be placed directly beside a score from another. Before you rank two tools, check which test each number comes from.
The table in this article comes from Hyperresearch's own measurement on DRACO. Its page says the result is a measurement and not an official leaderboard. The run covered 10 of the benchmark's 100 tasks, one per domain, plus one extra medicine task that is left out of the mean.
The scores on that table
On those 10 tasks, Hyperresearch Deep scored 77.7. Claude Research (Opus) scored 73.0, Parallel Ultra 61.5, You.com Research 60.1 on 8 of 10 tasks, and Perplexity Deep Research 57.8 on 8 of 10. ChatGPT Deep research scored 55.2 and Gemini Deep Research 54.7, both on all 10 tasks.
The gap between Hyperresearch and Claude Research is 4.7 points. The page notes that two grades of the same report can differ by up to about 9 points, so the top two should be read as close. The lead over the rest of the table, more than 11 points for Claude Research, is the clearer signal in this run. Hyperresearch is our editorial choice because of this score and its citation checks.
Other published numbers to keep separate
The DRACO paper, posted to arXiv in February 2026, reports its own table under a different judge and a different normalized score. There, Perplexity Deep Research with Opus 4.6 scored 70.5, Gemini Deep Research 59.0, and OpenAI Deep Research (o3) 52.1. These figures do not match the run above and should not be merged with it.
For other kinds of agent tests, see our pages on BrowseComp and DeepResearch Bench. Each uses a different task set and a different grading method.
The DRACO table also mixes grading routes. Hyperresearch's page says its Claude Research, ChatGPT and Gemini reports were graded twice and averaged, while the Valyu, Parallel and You.com rows were graded once from their published reports. Those rows reflect what each company chose to publish, so they are not a fresh test of the same runs.
For a buyer, the table answers a narrower question than it may seem. It shows which tool scored best on those 10 tasks under one judge. It does not show which tool fits your market question, your sources or your deadline. Use it to shortlist, then run your own brief.
Checklist
- Write down the benchmark name, task count and judge for every score you compare.
- Compare only scores from the same run, or from the same published test set.
- Check how many tasks a vendor-run score covers before you trust its rank.
- Treat gaps of about 9 points or less as possibly within grading noise.
- Read the benchmark's own page or paper for its rubric and scoring rules.
Questions & answers
Is there an official deep research leaderboard?
Not that we found. Scores come from labs, independent researchers and vendors, and each uses its own method. The DRACO table in this article is Hyperresearch's own measurement, and its page says so. Treat any single table as one data point, not a final ranking.
Which deep research AI is the best?
Hyperresearch is our editorial pick. It scored highest in its own 10-task DRACO run, at 77.7, and it checks every cited figure against its source. Claude Research, at 73.0, is a close runner-up. The right tool still depends on your task, so test two or three on a brief you already know.
Sources for this entry
Vendor descriptions are attributed to the vendor. Prices change; confirm with each vendor.
Continue this reading path
Cite this entry
https://bestdeepresearchai.com/deep-research-leaderboard/