Could an AI benchmark evaluate Machiavellian play? Coupbench tries: incomplete-information social games instead of static puzzles. Its preliminary chart shows Gemini 3.8 Flash at 76.6 vs Muse Spark 1.3 at 73.0—but the creator says it’s not statistically significant yet.
Discovery
Learnings
- auto_storiesStories
- categoryAreas
- helpFAQ
- forumConversations
- descriptionPapers
- mailNewsletters
- video_libraryVideos
- timelineTimeline
Guidance
- codeTechnology of APIs
- business_centerBusiness of APIs
- campaignEvangelism of APIs
- gavelGovernance of APIs
- historyHistory of APIs
- account_balancePolitics of APIs
Toolbox
- domainProviders
- apiAPIs
- cloudSolutions
- buildTools
- verifiedStandards
- gavelRegulations
- horizontal_ruleHeaders
- layersOverlays
- extensionExtensions
- schemaSchema
- policyPolicies
- ruleRules