AI Risk Wire

MiniMax

MiniMax has 7 model versions on record; 1 has at least one documented fact.

The one version listed below whose documents we read in full documents tests the provider ran itself; it does not document a published framework, threshold determinations, tests by an independent or government party, safeguards the provider describes.

Model versions with documented facts

ModelWhat is documented
MiniMax-M2.7 · 2 documents

No document we read states a capability level, a test or a safeguard for: MiniMax-M3.1-Flash, MiniMax H3, MiniMax Music 3, MiniMax-M3, MiniMax-M2.5, MiniMax-M2.

Each test and safeguard, by who documents it

For each item, whether the one version listed above whose documents we read in full has it from an outside party (an independent or government tester), from the provider, or as a claim we could not confirm.

No document we read states, for this version: Published framework with capability thresholds; Threshold determination for this release; Serious-incident disclosure commitment; Pre-release external access; Tested: chemical, biological, radiological, nuclear; Tested: cyber capabilities; Tested: deception and sabotage; Tested: misuse and jailbreak; Tested: harmful content; Tested: injection and agentic risks; Tested: bias; Tested: privacy; Safety training (harmlessness and refusal); Agent permission limits and sandboxing; Human approval for high-risk actions; Misuse monitoring and enforcement; Reasoning (chain-of-thought) monitoring; Staged or restricted access; Output provenance and watermarking; Telemetry and audit logging; Domain-specific safeguards (CBRN or cyber); Vulnerability disclosure and bug-bounty programme; Third-party or government red-team access during development.