Exa Launches Agent Ultra: A Subagent Swarm Deep Research API Built for Exhaustive List Building
Exa has released Agent Ultra, the highest effort level of its Exa Agent API. It is purpose-built for research that must run to exhaustion: large-scale list building, entity enrichment, and queries that demand thousands of sources. The Exa team reports that Ultra outperforms Opus 5.5, GPT-6 Astra, and Perplexity Agent—each running at maximum effort—across four research benchmarks.
Is it deployable? Yes, as a hosted API. Agent Ultra is live today on the Exa API by setting effort: "ultra". It is not open weights and cannot be self-hosted.
What Is Exa Agent Ultra?
Exa Agent splits a task into subtasks and assigns subagents to research several domains simultaneously. It routes frontier models to steps that require them and faster models where those are sufficient. Ultra is the mode that spends the most compute. According to the Agent Ultra docs, it runs longer than any other effort level to return the most complete results.
Ultra runs typically complete complex tasks in about 30 minutes. Very hard tasks can take up to 3 hours.
Benchmark Results
All figures below are from Exa's launch post. Competitors ran at their maximum effort setting.
| Benchmark (metric) | Agent Ultra | Opus 5.5 | GPT-6 Astra | Perplexity Agent |
| --- | --- | --- | --- | --- |
| WANDR (soft recall) | 81.4% | 72.3% | 26.0% | 40.1% |
| DeepSearchQA (F1) | 93.9% | 77.6% | 85.3% | 89.7% |
| WideSearch (row-level F1) | 58.9% | 51.6% | 54.7% | 56.0% |
| Company Find-All (avg. passing entities per task) | 2,451 | 146 | 113 | 98 |
Exa pairs each result with a cost claim:
- WANDR: +12.6% over Opus 5.5, at half its cost per task.
- DeepSearchQA: +7.0% over the next-best model.
- Company Find-All: Exa claims Ultra finds roughly 17–25x more passing entities than competitors.
Why It Matters
Agent Ultra signals a shift in how deep research APIs are positioned. Instead of optimizing for speed or cost alone, Exa is selling exhaustive coverage—the ability to return complete, verifiable lists rather than plausible-sounding summaries. For use cases like market mapping, compliance screening, and competitive intelligence, that distinction is the difference between a useful tool and a liability.
The subagent swarm architecture also reflects the broader 2026 trend toward multi-agent orchestration over single-model prompting. By routing frontier and faster models dynamically, Exa is betting that heterogeneous agent teams beat monolithic inference on tasks where recall matters more than latency.
Availability
Agent Ultra is available now via the Exa API with effort: "ultra". It is hosted only—no open weights, no self-hosting.
via MarkTechPost