What Is The Maximum Number Of Agents MiroFish Can Simulate Simultaneously?

MiroFish is architected to handle up to one million simultaneous autonomous agents in a single simulation run. Each agent operates with its own personality profile, long-term memory, and behavioral logic derived from the seed material you provide.

Which LLM Models Are Compatible With MiroFish?

MiroFish supports any LLM that exposes an OpenAI-compatible API. The official recommendation is Alibaba's Qwen-plus model via the Bailian platform, though models like GPT-4o, Claude, and locally hosted alternatives work equally well by updating the environment variables.

How Does MiroFish Build Its Simulation World From A Document?

Upon receiving seed material, MiroFish runs a five-stage pipeline: it extracts entities and relationships into a knowledge graph via GraphRAG, generates agent personas from that structure, runs dual-platform parallel simulation rounds, updates temporal agent memories after each round, and finally synthesizes a ReportAgent-driven prediction output.

Can I Continue Querying The Simulated World After The Report Is Generated?

Yes. The Deep Interaction layer lets you chat directly with any individual agent inside the simulated world or interrogate the ReportAgent with follow-up questions — effectively treating the generated scenario as a persistent, queryable environment rather than a one-shot output.

What File Types Does MiroFish Accept As Seed Material?

MiroFish accepts PDF, Markdown (.md), and plain text (.txt) files as seed inputs. You can also start with a plain-language typed prompt if you do not have a structured document ready, and add files later to ground the simulation in specific source context.

Listed · Reviewed Jun 2026 · 7-criteria rubric

AI Business Tools

MiroFish

Name: MiroFish
Availability: InStock
Author: MiroFish (Incubated by Shanda Group)

Swarm Intelligence Engine That Rehearses The Future With AI Agents

by MiroFish (Incubated by Shanda Group) · Beijing, China · Founded 2025

Visit MiroFish →

Pricing

From $90/year

What is MiroFish?

MiroFish is an open-source, multi-agent swarm intelligence prediction engine that converts real-world seed material — policy drafts, market reports, business plans, or story fragments — into a fully interactive digital simulation.

It spawns thousands of AI agents, each with distinct personalities, long-term memory, and behavioral logic, to model how populations react, amplify, or resist a given scenario.

Built on CAMEL-AI's OASIS simulation framework and backed by China's Shanda Group, MiroFish outputs structured prediction reports with risk signals, narrative paths, and follow-up questions — giving analysts, strategists, and researchers a decision rehearsal sandbox that goes far beyond static forecasting.

MiroFish — Swarm Intelligence Engine That Rehearses The Future With AI Agents Whether you're evaluating MiroFish for your team or comparing it to alternatives in the AI Business Tools category, this in-depth review covers everything: features, pricing, real user reviews, pros and cons, integrations, and direct comparisons against competitors.

Key Features 8

Upload Any Report to Instantly Simulate Future Social Scenarios

1M+ Parallel AI Agents per Run for Swarm Intelligence Simulation

Auto Ontology Generation from Raw Reports, Notes, or Fiction

Living Relationship Graph Builder Mapping Actors, Tensions, and Memory

Dual-Surface Simulation Across Twitter and Reddit Style Channels

AI-Generated Prediction Reports with Risks, Confidence Signals, and Key Actors

Deep Interaction via ReportAgent or Direct Character Interviews

Multi-Use Case Coverage: Public Opinion, Policy, Brand Crisis, and Fiction

Who Is MiroFish For

1 Policy Analysts And Government Strategists

2 Marketing And Brand Campaign Teams

3 Financial Risk Researchers

4 Business Strategists And Management Consultants

5 Academic Researchers In Social Simulation

Pros & Cons

Pros 4 benefits

Massive Agent Scale (1M)
Fully Open-Source Codebase
LLM-Agnostic Architecture
Backed By Shanda Group

Cons 3 limitations

High LLM API Consumption Costs
Complex Self-Hosted Setup
No Native GUI Mobile Support

Frequently Asked Questions

5 questions

Who is MiroFish for?

MiroFish is most useful for Policy Analysts And Government Strategists, Marketing And Brand Campaign Teams, Financial Risk Researchers and Business Strategists And Management Consultants.

MiroFish pricing

MiroFish uses a freemium model: a usable free tier with optional paid upgrades. Headline pricing: From $90/year. For the current tier breakdown and any limits, see the pricing section above or check the vendor's pricing page directly — limits and prices change.

What's New

Backend README Technical Documentation Update N/A

Comprehensive update to the MiroFish backend technical documentation, clarifying deployment configurations and API environment variable setup.

Dec 2

MiroFish Cloud Platform Blog Launch N/A

Official cloud-hosted platform at mirofish.my went live with tiered annual subscription plans, a blog, and a full web-based simulation console.

May 1

View all updates

User Base

17K+ GitHub Stars

Active Users

Security & Privacy

Not Publicly Specified

GDPR Compliant (UK GDPR)

Data Processed Only For Service Delivery Third-Party Infrastructure Limited To Hosting, Auth, And Payments User Data Deletion And Export Rights Honored On Request

Collaboration & Teams

Comments

Learning & Support

Resources

Documentation Video Tutorials Blog

Community

Discord

Support Channels

Email Priority

Localization

UI Languages

Content Languages

Multi-Language UI

Recognition & Trust

VC Funded Open Source

Media: Featured In Medium, Dev.to, LinkedIn Pulse, 36Kr English Edition

All Features of MiroFish