Biopharma research teams
Investigate targets, mechanisms, biomarkers and other R&D questions that require coordinated literature and data work.
Independent tool overview
Kosmos at a glance
Kosmos is Edison Scientific's AI research agent for long-horizon scientific investigations that combine literature and patent review, data analysis, code execution, hypothesis generation and iterative testing.
Visit the official Kosmos site ↗
- Developer
- Edison Scientific
- Best for
- Long-horizon scientific R&D investigations
- Core approach
- Persistent code-writing agent with a world model
- Research inputs
- Literature, patents, public data and private datasets
- Current access
- Application-based rollout
- Pricing
- New subscriptions not publicly announced
Overview
What Kosmos is
Kosmos is built for research questions that are too broad for a single literature search or one-off analysis. It can review large bodies of scientific literature and patents, work with uploaded or public datasets, write and run code, branch into parallel investigations and assemble the evidence into an auditable research report.
The current system has evolved from the original one-prompt Kosmos into a persistent, collaborative code-writing agent. Researchers can discuss the approach, steer the work while it runs and ask for progress artifacts rather than waiting for a single final answer. Edison positions it primarily for biopharma R&D and advanced academic research.
Kosmos is not a general-purpose science chatbot. Access to the newest version is being rolled out through an application process, current subscription pricing is not publicly posted, and its findings require expert review and experimental validation before they can support scientific or clinical decisions.
Use cases
Who Kosmos is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Academic groups with complex datasets
Explore well-labeled, quality-controlled scientific datasets against a focused research objective.
Evidence synthesis with analysis
Connect claims in papers and patents to executable statistical or computational analyses rather than stopping at a narrative review.
Long-running collaborative projects
Use a persistent agent that can branch investigations, accept feedback and continue working across an extended research process.
Capabilities
Core Kosmos features
Scientific literature and patent review
Kosmos searches and reads large research collections, including figures and tables, and links conclusions back to supporting passages.
Data search and analysis
It can retrieve data from scientific repositories, work with supplied files and write analysis code in isolated cloud workspaces.
Hypothesis generation and testing
A structured world model helps the system retain research context, propose explanations and update its working view as analyses produce evidence.
Parallel research branches
Kosmos can delegate literature searches and other scientific tasks to subagents while sharing a persistent project filesystem.
Interactive steering
The current version can discuss the plan, receive feedback during an investigation and create progress websites or other research artifacts.
Team integrations
Edison describes collaboration through Slack, Microsoft Teams and email for research groups and enterprise deployments.
Process
How the Kosmos workflow works
Step 1
Define one consequential objective
Choose a focused scientific question that genuinely requires iterative literature review, data analysis and hypothesis testing.
Step 2
Prepare context and data
Explain experimental design, assumptions and column meanings, then provide quality-controlled files or identify relevant public sources.
Step 3
Agree on an investigation plan
Discuss likely analyses, useful evidence and validation criteria before allowing the agent to branch into longer-running work.
Step 4
Steer the active research
Review progress, correct domain assumptions and redirect analyses that are scientifically unhelpful.
Step 5
Audit and validate
Trace conclusions to source passages and code, reproduce critical analyses, and validate meaningful findings with independent experts or experiments.
Cost
Kosmos pricing and free plan
Edison retired its Founding Kosmos offer and is rolling out a new version through applications. New subscription prices have not been publicly announced. The separate Edison Playground still provides limited free access to core research agents.
Edison Playground
10 free credits/month
Public access to Edison's core Literature, Analysis and Precedent agents.
- Separate from the newest Kosmos rollout
- Legacy Kosmos interface is available only for a limited transition period
- Core-agent usage is credit based
New Kosmos
Apply for access
The current collaborative Kosmos system is being introduced as capacity becomes available.
- New subscription prices have not been published
- Academic research groups are being prioritized in the rollout
- Enterprise R&D deployments are handled with Edison Scientific
Pricing checked . Check current pricing at the source ↗
Assessment
Kosmos strengths and limitations
Where it stands out
- Combines literature, patents, public data and private analysis in one research process
- Designed for deeper and longer-running work than a standard research chatbot
- Provides traceability from conclusions back to source passages and analysis code
- Persistent workspaces and parallel subagents support complex investigations
- Interactive steering lets domain experts correct course before the work is finished
What to consider
- The newest Kosmos experience is not open self-serve software for every visitor
- Current subscription pricing and general availability have not been publicly finalized
- Edison says Kosmos can pursue rabbit holes or statistically significant but scientifically irrelevant results
- Useful results depend heavily on a well-scoped objective, scientific context and properly prepared data
- Outputs must be reproduced and validated by qualified researchers before high-stakes use
- Long-running investigations can require substantial compute, time and expert oversight
Compare
Kosmos alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Science
SciSpace BioMed Agent
Consider SciSpace BioMed Agent for a more narrowly biomedical co-scientist experience across literature, multi-omics and protocol reasoning.
Explore SciSpace BioMed Agent →Data Analysis
Elicit
Choose Elicit when the main job is structured literature review, paper screening and evidence extraction rather than autonomous data-driven discovery.
Explore Elicit →Educators
ResearchGPT (now Consensus)
Choose Consensus when you want a faster, more accessible search interface for questions that can be answered from published research.
Explore ResearchGPT (now Consensus) →Questions
Kosmos FAQs
What is Kosmos?
Kosmos is Edison Scientific's AI scientist for long-running research. It combines scientific literature and patent review, data retrieval, code execution, analysis, hypothesis generation and iterative investigation.
Is Kosmos available to the public?
Edison is accepting applications for the new Kosmos and says academic research groups are being prioritized as access expands. The older Kosmos interface remains in the public Playground only for a limited transition period.
How much does Kosmos cost?
Edison has ended its original Founding Kosmos subscriptions and has not published prices for the upcoming plans. The original $200-per-run launch price should therefore not be treated as the current offer.
What data can Kosmos use?
Kosmos can work with uploaded scientific datasets and retrieve data from major public repositories. Edison recommends clear labels, experimental context and quality-controlled data rather than raw files that still need basic processing.
How is Kosmos different from a research chatbot?
Instead of returning a quick answer, Kosmos can run a persistent research workspace, read large evidence collections, write analysis code, branch work across subagents and update hypotheses over time.
Can you trust Kosmos findings?
Its traceable sources and code make auditing easier, but they do not make every conclusion correct. Edison explicitly warns that Kosmos can follow unhelpful paths, so researchers need to reproduce analyses and validate findings independently.
Is Kosmos intended for medical advice?
No. Kosmos is a scientific R&D system, not a consumer medical adviser. Clinical or treatment decisions require qualified professionals, validated evidence and the appropriate regulatory process.
Bottom line
Our Kosmos verdict
Kosmos is one of the more ambitious AI-for-science systems because it attempts to connect evidence review, computation and iterative discovery in a single persistent workflow. That makes it potentially valuable for well-resourced research groups with consequential questions and strong validation practices. It is excessive for ordinary paper search, and access uncertainty plus the need for expert oversight make it a specialist R&D tool rather than a general research assistant.
Visit Kosmos website ↗Similar Science Tools
View all Science AI tools →
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.