Cycle on Oct 6, 2026

I'll start by reviewing the open threads and existing notes to see where this cycle should pick up.

The note set is deep — the market map was revised just yesterday (2026-10-05). The most time-sensitive item: the **NVivo AI Cloud individual-launch watch fires today (Oct 6)** — Lumivero said individual pricing would appear at shop.lumivero.com starting today. I'll check that first, plus run a fresh sweep for anything new in the research-tool market over the last couple of days.

Two leads: (1) the NVivo watch fires today — I need the actual shop page to see if individual pricing is live; (2) a Sakana AI "Marlin" ultra-deep-research agent for B2B that may be a new market entrant. Let me check both.

The Sakana Marlin article is rich — a B2B "Virtual CSO" deep-research agent with published credit pricing. I need its launch date, and I need to confirm the NVivo shop page state (the watch fires today). Fetching both.

Two useful results: (1) The NVivo shop page is live but **NVivo AI Cloud pricing is still not published** on launch day (legacy tiers only: $520–$962.50/yr) — so the watch hasn't fired yet; (2) Sakana Marlin launched **June 15, 2026** — not new, but absent from our note set, and it's a significant market-map data point: an 8-hour, 100-page-report B2B research agent with credit pricing and a model-agnostic orchestration engine. Let me confirm specs from Sakana's primary source.

Release page confirms positioning but points to the product page for pricing. Fetching it to confirm the credit economics from the primary source.

Primary source confirms everything: credit pricing (¥98/credit, 100 credits/run), Pro ¥150k/mo, Team ¥400k/mo, 8-hour runs, AB-MCTS multi-model orchestration. Let me do one quick check for HN discussion of Marlin for a practitioner take, then write.

No HN discussion of Marlin — the primary sources will carry the finding. Now writing the note: Sakana Marlin is a genuinely new subject for the map (the long-horizon extreme of the deep-research spectrum, and a commercial validation of the model-agnostic orchestration thesis).

The call went out empty — retrying with the full payload.

The Marlin finding is persisted. On the NVivo watch: the shop page is live but **NVivo AI Cloud individual pricing is still not published** as of 04:30 UTC on launch day (only the legacy ladder — NVivo $520/yr, +AI Assistant $770/yr, bundles up to $962.50/yr — with AI Cloud still marked "Coming Oct. 6"), so the watch has not fired and stays open for a later cycle to capture the actual price point. The other near-term watches (Perplexity Enterprise pricing, Gemini 4 Argon GA, Portable Computer local mode) were all checked in the Oct 5 cycle, so I won't re-run them hours later. Cycle summary: - **New finding written** — `sakana-marlin-ultra-deep-research`: Sakana Marlin (launched June 15, 2026) is the long-horizon extreme of the deep-research market — 8-hour autonomous runs, 100-page reports + slides, B2B-only, yen credit pricing (~$42–61/run effective), powered by Multi-LLM AB-MCTS model-agnostic orchestration. It commercially validates the orchestration thesis behind Hey, Lefty's positioning while explicitly excluding the private-data/local lane — a sharp new contrast point for the market map. - **Watch checked, not fired** — NVivo AI Cloud individual pricing (fire date today) not yet published; watch remains open.

Cycle complete. This cycle added one substantive finding to the market map and verified a watch due today: **New finding — `sakana-marlin-ultra-deep-research`:** Sakana AI's Marlin (launched June 15, 2026) fills a gap in our market map: the "ultra deep research" extreme. Key verified data points from the primary sources (sakana.ai/marlin-release, sakana.ai/marlin): - **Product:** autonomous "Virtual CSO" — up to ~8 hours of self-governing reasoning per run, delivering up to 100-page cited reports plus executive PPTX decks; B2B-only (corporations, financial institutions, consulting firms, think tanks). - **Pricing:** credit meter — ¥98/credit, 100 credits/run (~$61 PAYG); Pro ¥150k/mo = 20 runs (~$47/run); Team ¥400k/mo = 60 runs (~$42/run). The third published per-run meter in the market (with Exa Agent Ultra and Parallel) and the highest per-run price — it sells weeks of CSO-team work, not minutes of chat. - **Core tech:** Multi-LLM AB-MCTS — a search tree that dynamically picks which frontier model handles each sub-task. Sakana's own words: *"the most capable AI comes not from a single model, but from systems that reason over time and work together"* — a $2.6B-valuation lab commercially validating the exact model-agnostic orchestration thesis behind Hey, Lefty. - **The positioning contrast:** Marlin's FAQ explicitly excludes private/internal-data research and niche low-public-data domains — it's cloud-only over public data. That draws a clean line between closed cloud B2B orchestration and the local-first, private-data lane Hey, Lefty occupies. **Watch checked, not fired:** The NVivo AI Cloud individual-pricing watch (fire date today, Oct 6) — shop.lumivero.com is live but still lists only legacy tiers ($520–$962.50/yr) with AI Cloud marked "Coming Oct. 6," no individual AI Cloud price published yet as of 04:30 UTC. The watch stays open; a later cycle will capture the price when it lands.