By MiroMindAI
MiroThinker
MiroThinker is a deep research agent optimized for complex research and prediction tasks. Our latest models, MiroThinker-1.7, achieves 74.0 and 75.3 on the BrowseComp and BrowseComp Zh, respectively.
Filed under Agent Frameworks. Status: Published
- Stars
- 8.2k
- Forks
- 634
- Open issues
- 1
- Last commit
- 1mo ago
- Stats refreshed
- Refreshed May 22, 2026
On the maker
Use cases
- Tool Use / Function CallingTools that support OpenAI-style function calling, MCP tool use, or structured tool dispatch.
Pricing
Open-source
Field notes
No field notes yet.
Field notes for MiroThinker will land here when sources support a confident take ; synthesized from postmortems, vendor retros, dev-team blogs, deeply-engaged GitHub issues, and our own builds.
Coverage isn’t promised on every tool ; empty sections are honest. Field notes are curated, not generated from vendor copy.
Benchmarks
Scores aren’t in yet.
We’re wiring up SWE-bench, Aider Polyglot, and a custom dev-task suite next. Methodology will be public; vendor pre-notification is 48 hours.
View benchmarksHow we make money
This directory is supported by display advertising. Advertisers do not influence editorial rankings, benchmark scoring, or which tools are featured. Tools are ordered by data.