MiroThinker
Deep research agent and open models tuned for complex research and forecasting tasks.
Updated
✨ Try AgentScout's free AI image generator →Good for: research
Use cases
About MiroThinker
MiroThinker is listed in our models & local runtimes category and is aimed at builders. Deep research agent and open models tuned for complex research and forecasting tasks. Pricing, as the maker states it: Open weights. Maturity: Mature. We last checked the official link on 2026-10-16.
What is MiroThinker?
MiroThinker is a deep research agent and open models optimized for research and prediction.
Its README reports a score of 88.2 on the BrowseComp benchmark; MiroThinker-1.7 and 1.7-mini were released in March 2026.
Sources: the official README and our directory listing, checked 2026-10-18. Official site: miromind.ai.
Who is MiroThinker for?
Researchers and builders working on deep-research agents.
MiroThinker at a glance
- Open source: Yes (Apache License 2.0)
- GitHub: MiroMindAI/MiroThinker · 8,423 stars (as of Oct 21, 2026)
Key features
- Deep research agent
- Open models
- Forecasting tasks
MiroThinker FAQ
What is MiroThinker?
MiroThinker is a deep research agent and open models optimized for research and prediction.
Who is MiroThinker for?
Researchers and builders working on deep-research agents.
Is MiroThinker free?
MiroThinker's pricing, as the maker states it: Open weights. Check the official site for current plans.
Is MiroThinker open source?
Yes. MiroThinker is released under the Apache License 2.0, according to its GitHub repository.
What are alternatives to MiroThinker?
Similar tools include LocalAI, Ollama, GLM-5, Meta Muse Wonder, Jan and Oumi.
MiroThinker ratings and reviews
Be the first to rate MiroThinker: tap 1 to 5 stars below. A written review is optional.
MiroThinker alternatives
Other models & local runtimes tools in our directory. See all models & local runtimes tools →
LocalAI
Self-hosted engine with an OpenAI-compatible API for running LLM, vision, voice and image models on your own hardware, including agents.
Ollama
Runs open models such as Qwen, Gemma, DeepSeek and gpt-oss locally or in Ollama's cloud, and connects them to agents like Claude Code, OpenClaw and Codex.
GLM-5
Z.ai's open-weights GLM-5 model family built for coding and long-horizon agentic tasks, also offered through the Z.ai API.
Meta Muse Wonder
Meta's new models for personal and professional agent tasks, available via Meta Model API and Muse Code.
Jan
Open-source ChatGPT alternative that downloads and runs LLMs fully offline on your computer, with optional connections to cloud models.
Oumi
Open platform for fine-tuning, evaluating and deploying open agentic LLMs and VLMs.
Related lists
🧭 Task guides: AI research and report agents
💲 Estimate API cost: compare per-request and monthly costs for GPT, Claude, Gemini and Grok models.
Badge for makers
Built MiroThinker? Add this badge to your README or site. It links back to this page. About badges
