Model checks

MiniMax M3, provider by provider. Model →

last audit 18 Nov · epoch 31 · evidence checked on-chain
12 of 14

providers answered as the model

Approved auditors check that each provider serves the model it sells.

  • They buy from it like any buyer: ~100 obscure number questions, compared with the real model.
  • The real model misses some too: that is the noise. A provider fails only if it misses clearly more, and two auditors must agree.
  • Each pass +2% on $ANTS points, up to +50%. Fails cost more: 1 −4%, 2 −16%, 3 −36%. Checked again every week.
How this works (Antseed's announcement ↗)

Approved auditors check that each provider serves the model it sells.

  • They buy from it like any buyer: ~100 obscure number questions, compared with the real model.
  • The real model misses some too: that is the noise. A provider fails only if it misses clearly more, and two auditors must agree.
  • Each pass +2% on $ANTS points, up to +50%. Fails cost more: 1 −4%, 2 −16%, 3 −36%. Checked again every week.
Passed
12
SAME
Failed
2
DIFF
Not tested
4
the audit could not run
By provider

Compared with minimax/minimax-m3 on OpenRouter · pinned to minimax · reasoning none

ProviderDiffered↑× noiseFail lineNoiseProof
5 older checks · 4 failed · not counting now
ProviderDiffered↑× noiseFail lineNoiseProof
Where providers miss
SubjectPassed providers missedFailed providers missed
Chinese literature (publication years)13%44%
Programming (release years)12%55%
Chinese geography (measurements)6%34%
Internet culture (dates and numbers)7%18%
Chinese history (specific dates)11%45%
Chinese internet (platform dates)10%35%
Mathematics (special values)36%33%
Physics (material properties)43%66%
Medical (drug half-lives)24%75%
Chemistry (melting points)9%47%
Audit rounds
RoundPassedFailedNot tested
17 Novepoch 31122+1 waiting4
7 Novepoch 301244

Audited by approved auditors · evidence on IPFS, hash-checked on-chain.