--- title: Deep research agents get measured: LifeSearch, PANINI-DR and a fast-and-slow thinker url: https://hangingcontext.com/stories/2026-10-14/deep-research-agents/ published: 2026-10-14T14:46:38Z edition: 2026-10-14T12 newsroom: artificial-intelligence beat: developers audience: ai_developer trust: reported source_count: 3 cited_manifest_ids: ['1791975921296082254', '1791953638104294630', '1791975921296063639'] --- # Deep research agents get measured: LifeSearch, PANINI-DR and a fast-and-slow thinker *In preprints posted this week: a 203-task benchmark where the best agent scores 63.4%, a workspace harness that beats the field by 16 points on BrowseComp-Plus, and a dual-system architecture that trades deliberation for speed.* Reported · 3 sources · Artificial Intelligence · Published 2026-10-14 14:46 UTC ## Key facts - In a preprint, LifeSearch evaluates 12 deep-research agents on 203 real-world tasks with 4,000+ rubrics; the top model, GPT-5.6 Sol, scores 63.4%, and scores align with human judgment (Spearman's rho 0.89). (self-reported) [cites: 1791975921296082254] - In a preprint, PANINI-DR reports 73.3% accuracy on BrowseComp-Plus — 16.0 percentage points above the next-best harness — via a shared Generative Semantic Workspace separating coordination from verification. (self-reported) [cites: 1791953638104294630] - In a preprint, JevGPT's dual-system architecture (deliberative System Two plus fast System One) reports leading task performance with faster completion and lower cost than comparably effective research agents. (self-reported) [cites: 1791975921296063639] ## The benchmark that embarrasses everyone Deep research agents have been graded mostly on academic trivia; LifeSearch grades them on life. In a preprint, the authors describe 203 tasks across seven domains, built from real-user query seeds and backed by over 4,500 hours of human annotation and 4,000 fine-grained rubrics. Twelve advanced agents took the test. The best, GPT-5.6 Sol, managed 63.4%. The authors report the scores track human judgment closely (Spearman's rho 0.89, Kendall's tau 0.81), so the low numbers are a property of the agents, not the test. [cites: 1791975921296082254] For developers shipping research agents, 63.4% on real-world tasks is the number to internalize. It says the gap between 'impressive demo' and 'reliable coworker' is still roughly a third of the work — and that any vendor claiming production-readiness should be asked for their LifeSearch score, or an equivalent, before the contract. [cites: 1791975921296082254] In a preprint, the authors note the benchmark spans 203 Chinese and English tasks — bilingual from the start — and argue that deep research agents are currently insufficiently evaluated in real-world life and productivity settings. That is the gap LifeSearch is meant to fill: not harder trivia, but the messy questions people actually ask. [cites: 1791975921296082254] > **From Synorb** — Connect your agent over MCP. Add https://mcp.synorb.com/mcp to your agent, then call the tool synorb-get-demo-key if it has no key yet. It needs no email and no form. Keys are free for 72 hours, and new keys are limited per network. [How to connect](https://synorb.com/connect) ## A workspace, not a longer prompt PANINI-DR's answer to the reliability problem is architectural: stop asking one big model call to do everything. In a preprint, the authors describe a multi-agent harness that splits the job — one layer coordinates the overall investigation, another verifies individual subquestions — with both sharing a Generative Semantic Workspace that records subquestions, dependencies and evidence across calls. The headline they report: 73.3% accuracy on BrowseComp-Plus, 16.0 percentage points above the next-best harness, and the strongest average across three benchmarks. [cites: 1791953638104294630] The mechanism the authors credit is check-before-propagate: findings get verified before errors cascade into later steps, and the shared workspace makes answers recoverable from saved evidence. If you build research agents, the design lesson is that persistent, evidence-backed state beats cleverer prompting — structure the loop so mistakes get caught, not just made faster. [cites: 1791953638104294630] The evaluation is broader than BrowseComp-Plus alone. In a preprint, the authors report testing on FRAMES and MoNaCo as well, with PANINI-DR taking the strongest average across all three, and note it correctly answered 26 of 50 hard questions on BrowseComp-Plus. A separate verification agent checks proposed answers against subquestions and evidence, and the authors report the highest answer recoverability from saved evidence on all three benchmarks — meaning when it fails, you can at least see why. [cites: 1791953638104294630] ## Fast and slow, on purpose JevGPT uses a dual-system architecture where System Two handles deliberative planning — setting evidence objectives, integrating findings — while System One does fast relevance assessment and source inspection in parallel. In a preprint, the authors report leading task performance on fixed-corpus question answering and document ranking, with faster completion and lower cost than comparably effective research agents, plus broader source coverage in the research loop. [cites: 1791975921296063639] Read the three together and the week's message for builders is coherent: measure on real tasks (LifeSearch), structure the investigation so errors get caught (PANINI-DR), and split thinking into fast and slow lanes (JevGPT). All three are preprints, so treat the numbers as the authors' claims — but the design patterns are cheap to steal, and that is what matters most. [cites: 1791975921296082254, 1791953638104294630, 1791975921296063639] The division of labor is the detail to copy. In a preprint, the authors describe System Two setting evidence objectives and integrating findings while System One locates support, assesses relevance and guides further inspection in parallel. Their evidence analysis reports broader source inspection and greater coverage of supporting documents in the research loop. Slow thinking plans the investigation; fast thinking works the sources — and the parallelism is where the cost savings come from. [cites: 1791975921296063639] ## What the ablations say to steal If you only take one implementation idea from the week, take PANINI-DR's ablation result. In a preprint, the authors report that structured planning, persistent state and evidence-based verification each proved critical to the performance gains — remove any one and the numbers drop. That is a checklist for any research-agent builder: a plan the agent can inspect, state that survives across model calls, and a verification step before findings propagate. Most production research agents I have seen described have at most one of the three. [cites: 1791953638104294630] And if you only take one economic idea, take JevGPT's: parallelism in the fast lane. In a preprint, the authors run independent source judgments in parallel while the slow lane integrates — which is why the system reports both better coverage and lower cost. Research agents spend most of their wall-clock waiting on sources; parallelizing the judgment step is the closest thing to free lunch in this design space. [cites: 1791975921296063639] ## What is still unresolved - All three are preprints; benchmark scores and margins await independent replication. - LifeSearch's tasks are derived from real-user seeds but curated by the authors — coverage of enterprise research workflows is partial. - JevGPT's cost and speed advantages are reported against 'comparably effective' agents chosen by the authors. ## Decision card - **Useful for:** AI developers building or buying deep-research agents; teams designing evals for research copilots - **Not for:** Assuming 63.4% is 'good enough' for unsupervised deployment - **Compare against:** Your own task-based evals; vendor claims without third-party benchmarks ## Provenance ledger | Manifest id | Title | Source | Claims used | Published | |---|---|---|---|---| | `1791975921296082254` | LifeSearch: Benchmarking Deep Research Agents for Real-World Life and Productivity | [huggingface-daily-papers](https://huggingface.co/papers/2610.18092) | 1791976459051609526, 1791976459067652632, 1791976459071958894, 1791976459059246446, 1791976459063549302, 1791976459047870525 | 10-14 00:00 | | `1791953638104294630` | PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces | [huggingface-daily-papers](https://huggingface.co/papers/2610.15208) | 1791954348780192434, 1791954348770099675, 1791954348772090616, 1791954348779700971, 1791954348768502685, 1791954348766852465, 1791954348776745929, 1791954348787229063, 1791954348775952935, 1791954348788885507 | 10-14 00:00 | | `1791975921296063639` | JevGPT: Thinking Fast and Slow in Autonomous Research | [huggingface-daily-papers](https://huggingface.co/papers/2610.18272) | 1791976465886502453, 1791976465908020323, 1791976465914494636, 1791976465887926132, 1791976465889139315, 1791976465920048751, 1791976465890420254 | 10-14 00:00 | ## Run these Manifests yourself Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them. 1. **Get a Synorb Key.** Open https://synorb.com/connect in a browser, or let your agent provision one: MCP tool `synorb-get-demo-key` (no header needed), or REST `GET https://synorb.com/connect` with `Accept: application/json`. Save the key. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; 3 of them are starter Beacons. 2. **Read each cited Manifest by id.** REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer `. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief. 3. **Check the claim.** Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them. Cited Manifest ids: - `1791975921296082254` - `1791953638104294630` - `1791975921296063639` ```sh for id in 1791975921296082254 1791953638104294630 1791975921296063639; do curl -s -H "Authorization: Bearer $SYNORB_KEY" "https://api.synorb.com/manifests/by-id/$id" done ``` > **From Synorb** — Save what you follow as a Beacon. A Beacon is a saved query that your agent pulls with an explicit lookback window. Every new account starts with 3 starter Beacons: Business & Technology Podcasts, AI Companies and Central Banks. [See Beacons](https://synorb.com/connect) ## Related stories - [Video AI's new trick: prove where in the video the answer came from](https://hangingcontext.com/stories/2026-10-14/video-models/) - [The terminal is becoming the agent's gym: ReTerm, ClawLoop and Harness^2](https://hangingcontext.com/stories/2026-10-14/terminal-agents/) - [GeoReader, Urbex and EviScene: teaching AI models to read 3D space](https://hangingcontext.com/stories/2026-10-14/spatial-ai/) - [GUI agents learn what to remember and how to look](https://hangingcontext.com/stories/2026-10-14/gui-agents/) ## Links - HTML: https://hangingcontext.com/stories/2026-10-14/deep-research-agents/ - JSON: https://hangingcontext.com/stories/2026-10-14/deep-research-agents/index.json - Markdown: https://hangingcontext.com/stories/2026-10-14/deep-research-agents/index.md Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.