--- title: "Sakana AI's Marlin Replaces Entire Research Teams With 8-Hour Autonomous Sessions" subtitle: "Sakana AI's first commercial product uses multi-model tree search to run autonomous strategy research sessions up to 8 hours long" url: "https://alphasignal.ai/news/sakana-ai-s-marlin-replaces-entire-research-teams-with-8-hour-autonomous" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "Sakana AI" company_url: "https://alphasignal.ai/company/sakanaailabs" published: "2026-06-15T02:48:31.000Z" updated: "2026-06-15T03:51:12.592Z" type: "News" topics: - "Reasoning" - "Business" subtopics: - "Test Time Compute" upvotes: 1214 read_time: "5 min" source_url: "https://sakana.ai/marlin-release/" comments: 0 --- # Sakana AI's Marlin Replaces Entire Research Teams With 8-Hour Autonomous Sessions > Sakana AI's first commercial product uses multi-model tree search to run autonomous strategy research sessions up to 8 hours long [Sakana Marlin](https://sakana.ai/marlin/) is the Tokyo-based AI lab's first commercial product, and it is not a chatbot. It is an autonomous research agent designed to sit down with a strategic question and not surface for up to eight hours, emerging with a structured slide deck and a multi-dozen-page report. No human in the loop. No follow-up prompts required. Sakana AI has launched a closed beta for Marlin, marking a notable evolution from its previous consumer-oriented Sakana Chat service. Marlin is designed to emulate the work of a CSO and an entire research team, compressing several weeks of strategic analysis into a single, unattended session lasting up to eight hours. The announcement also confirms a full self-serve commercial launch with tiered plans ranging from free pay-per-use up through Pro, Team, and Enterprise. ## Years of research, one product Marlin is not a wrapper around a single frontier model. It is built upon two core research developments Sakana has refined over the past two years. The first is Adaptive Branching Monte Carlo Tree Search (AB-MCTS), which conceptualizes reasoning as a tree-search problem, allowing the AI to autonomously determine which hypotheses to pursue, discard, or escalate to different models. This technique was recognized with a spotlight award at NeurIPS 2025. AB-MCTS is worth understanding in more detail. At its core, it reflects a philosophical shift in how we think about LLM reasoning. Rather than treating generation as a flat, linear process, Sakana's approach reframes inference as a strategic exploration through a search tree, navigating between depth (refinement of existing ideas), width (generation of new hypotheses), and even model selection itself. Think of it as the AI equivalent of a research team that knows when to go deeper on a promising lead versus when to pivot and explore a new angle entirely. The method uses a collection of LLMs to cooperate, perform trial-and-error, and leverage their individual strengths to solve complex problems. In benchmarks, the team used a combination of frontier models including o4-mini, Gemini 2.5 Pro, and DeepSeek-R1. The collective of models was able to find correct solutions for over 30% of the 120 ARC-AGI-2 test problems, a score that significantly outperformed any of the models working alone. ![Sample Sakana Marlin output showing strategy reports with charts, executive summaries, and multi-phase analysis documents](https://sakana.ai/assets/marlin-release/marlin-release-eng-figure.png) The second pillar is a workflow automation framework originating from the Sakana AI Scientist project, which demonstrated fully autonomous scientific discovery from ideation to peer review, and was published in Nature. Marlin essentially takes that scientific discovery pipeline and redirects it at business strategy problems. By combining these systems, Marlin is capable of executing hundreds to thousands of LLM queries per session while dynamically directing computational resources toward the most promising research avenues. That is a meaningful distinction from tools like Perplexity or even OpenAI's Deep Research, which typically complete in minutes. Marlin is designed for the kind of depth that takes hours. ## Who it is built for This focus on high-end research workflows positions Sakana AI to compete in knowledge-intensive segments such as consulting, corporate strategy, and financial analysis. The product is explicitly modeled around the idea of replacing a multi-week CSO-level research engagement. The target users are: - Financial institutions, corporate strategy and business planning departments, consulting firms, and think tanks. - Teams operating in fast-moving environments where traditional strategic research can take weeks, often rendering insights outdated by the time they are produced. - Organizations that need structured, sourced deliverables, not just chat summaries. ## Pricing and availability Sakana is going self-serve from day one. The launch includes a free pay-per-use tier alongside Pro, Team, and Enterprise plans, meaning teams can sign up and run a session without a sales call. During the beta period, participation is offered free of charge, and Sakana plans to refine the product based on user feedback ahead of broader commercialization. This is a significant strategic move. Sakana AI has operated primarily as a research lab since its founding, publishing work on evolutionary model merging, the AI Scientist, and AB-MCTS. The company is now valued at over $2.5 billion and is backed by NVIDIA, Google, and major Japanese financial groups. Marlin is the first signal that Sakana is ready to convert that research credibility into recurring revenue. ## The real question: does it hold up? The promise is compelling, but the risks are real. While Marlin is in beta, its real-world effectiveness remains under scrutiny. Businesses will need to weigh the benefits of speed against the potential loss of strategic depth. Automated reports at this scale introduce a hard-to-spot failure mode: confident-sounding analysis built on subtly wrong premises. Sakana's announcement does not directly address how Marlin handles hallucinations or source verification at scale. The competitive landscape is also heating up. Tools like OpenAI's Deep Research, Perplexity's research mode, and a growing number of agentic research products are all targeting similar workflows. What Marlin bets on is **duration and depth**: the willingness to run for eight hours and iterate through thousands of model calls rather than returning a quick synthesis. Whether that translates to meaningfully better strategy outputs is exactly what the beta is designed to test. If Sakana Marlin proves effective at compressing weeks of professional research into automated workflows, the product could eventually translate into a scalable SaaS-like revenue opportunity across multiple information-heavy industries. For a lab that has spent two years building the underlying technology, the commercial bet is now on the table. ## Sources - [X post by @sakanaailabs](https://x.com/sakanaailabs/status/2066352122183168004) - [Sakana AI Launches 'Marlin,' An Autonomous Research Assistant Built For The C-Suite](https://mpost.io/sakana-ai-launches-marlin-an-autonomous-research-assistant-built-for-the-c-suite/) - [Sakana AI launches 'Ultra Deep Research' to automate weeks of strategy work](https://the-decoder.com/sakana-ai-launches-ultra-deep-research-to-automate-weeks-of-strategy-work/) - [Sakana AI's TreeQuest: Deploy multi-model teams that outperform individual LLMs by 30%](https://venturebeat.com/ai/sakana-ais-treequest-deploy-multi-model-teams-that-outperform-individual-llms-by-30) - [How to build an AI scientist: first peer-reviewed paper spills the secrets](https://www.nature.com/articles/d41586-026-00899-w) - [Sakana AI Launches Closed Beta for Enterprise AI Research Assistant](https://www.tipranks.com/news/private-companies/sakana-ai-launches-closed-beta-for-enterprise-ai-research-assistant-2)