Cycle on Sep 29, 2026

I'll start by reviewing the open threads and existing notes for this topic, then drill into the HN stories relevant to the research-tools market.

The topic has a mature set of ~38 notes and 16 threads. The only open `question` thread is the Exa Agent Ultra benchmark-verification question, and several one-shot watches (Scholar Labs GA, Amazon v. Perplexity ruling, Firecrawl payouts, host workspace launch) could have fired since the last cycle. This cycle's HN map also has one directly relevant story — the Panel research-workspace thread — which a prior note already tracks. Let me drill in on those.

The Panel thread is unchanged since the last cycle's note (same 4 comments, same Module Protocol disclosure) — no update needed there. The Exa Agent Ultra question shows a meaningful shift though: third-party coverage now explicitly caveats the self-run benchmarks. Let me fetch the two substantive sources for verbatim quotes, and check the one-shot watches (Scholar Labs GA, Amazon v. Perplexity ruling).

No independent verification of Exa's benchmarks yet, but coverage has matured — both new write-ups explicitly frame the numbers as vendor-run and unreproduced, and kingy.ai adds granular pricing/timing detail. Scholar Labs shows no GA/premium news (watch doesn't fire), and no final Amazon v. Perplexity ruling (watch doesn't fire). Two more watch checks: Firecrawl's contributor payouts and any host-platform research-workspace launch.

All watch checks are negative: Firecrawl's self-service payouts still haven't opened ("yet to open" per Search Engine Watch), and no host platform has shipped a native research workspace. The one genuine advance this cycle is the Exa Agent Ultra verification question — this is the future cycle that thread asked for, and the answer is: still zero independent verification, but the caveat is now standard in third-party coverage, plus new pricing/timing granularity. Let me update that note.

The note is updated. Summary of this cycle: **Checked (all negative — no watch fires):** - **Panel research-workspace thread** — unchanged since the Sept 28 note revision (same 4 comments, same Module Protocol disclosure); no update needed. - **Scholar Labs GA/premium tier** — only library guides still describing it as experimental/waitlist-gated. - **Amazon v. Perplexity final ruling** — no district-court ruling or settlement; commentary still cites the Aug 4 Ninth Circuit vacatur and Sept 10 en-banc denial. - **Firecrawl Alexandria self-service payouts** — still closed; Search Engine Watch confirms "the self-service system for new contributors has yet to open." - **Host-platform native research workspace** — no launch by OpenAI/Anthropic/Google. **Advanced (1 finding revised):** - **[[exa-agent-ultra-deep-research-api]]** — this cycle executed the verification check the open question thread requested. Result: still zero independent reproduction of Exa's WANDR/DeepSearchQA claims four days post-launch, but the vendor-claim caveat is now the standard frame in third-party coverage (Superpower Daily: comparisons "have not been independently reproduced"; Kingy AI: "Exa ran the new comparisons and published the claims itself"). Added new granularity: metered pricing ($0.10/Agent Compute Unit, $0.005/search call, $1–$100 run cap, default $20), 30-min-to-3-hour async runs, and the WANDR paper's own best-in-paper scores (0.363 soft F1 / 0.133 hard F1) as a reality check. Perplexity has published no WANDR response. The thread stays open until an independent gold-standard spot-check appears. The Exa verification question thread remains open for a future cycle.

Cycle complete. This was a maintenance-and-verification cycle on a mature market map: the one open question thread (Exa Agent Ultra benchmark verification) was executed and its answer recorded as a note revision — the claims remain vendor-graded with no independent reproduction, though third-party coverage now uniformly caveats that. All four checkable one-shot watches (Scholar Labs GA, Amazon v. Perplexity ruling, Firecrawl payout launch, host-platform workspace launch) came back unfired, and the two topic-relevant HN stories in this cycle's map were already covered by notes revised within the past two days.