Model checks
Claude Fable 5.1, seller by seller. Model →
last audit 7 Nov · epoch 30 · evidence checked on-chain9 of 13
sellers answered as the model
How this works (Antseed's announcement ↗)
Approved auditors check that each seller 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 seller 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 seller 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 seller 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
9
SAME
Failed
4
DIFF
Not tested
5
the audit could not run
By seller
Compared with anthropic/claude-fable-5.1 on OpenRouter · pinned to anthropic · reasoning minimal
| Seller | Differed↑ | × noise | Fail line | Noise | Proof |
|---|---|---|---|---|---|
| Super SeederSAME | 5% | 1.0× | 19 under | 5.5% | ↗ |
| LLMartSAME | 6% | 1.2× | 18 under | 5.5% | ↗ |
| ▲ Apex AntSAME | 6% | 1.2× | 18 under | 5.5% | ↗ |
| Fire Ant 🔥🐜SAME | 7% | 1.3× | 17 under | 5.5% | ↗ |
| Open BirdSAME | 8% | 1.5× | 16 under | 5.5% | ↗ |
| 💬 BanterSAME | 8% | 1.5× | 16 under | 5.5% | ↗ |
| OpenCheapSAME | 10% | 1.8× | 14 under | 5.5% | ↗ |
| Venice.ai ProxySAME | 10% | 1.8× | 14 under | 5.5% | ↗ |
| Vikey SellerSAME | 12% | 2.1× | 11 under | 5.5% | ↗ |
| Inspectant 1 DIFF, the 2nd audit could not run | 43% | 7.8× | +22 over | 5.5% | ↗ |
| BestDIFF | 45% | 45× | +31 over | 1.0% | ↗ |
| Edith AIDIFF | 48% | 8.8× | +28 over | 5.5% | ↗ |
| Hana Gateway ✅DIFF | 51% | 9.3× | +31 over | 5.5% | ↗ |
| TokenSaleDIFF | 72% | 72× | +51 over | 1.0% | ↗ |
| incharge-austin not tested · seller unreachable | — | — | — | — | ↗ |
| attacompute.ai not tested · timed out | — | — | — | — | ↗ |
| Relay Works not tested · seller unreachable | — | — | — | — | ↗ |
| GesundAI not tested · seller unreachable | — | — | — | — | ↗ |
Where sellers miss
| Subject | Passed sellers missed | Failed sellers missed |
|---|---|---|
| Biology (chromosome numbers) | 13% | 51% |
| Programming (release years) | 10% | 53% |
| Medical (drug half-lives) | 3% | 54% |
| Physics (material properties) | 10% | 36% |
| Chemistry (melting points) | 16% | 57% |
| Chinese history (specific dates) | 11% | 34% |
| Chinese geography (measurements) | 2% | 44% |
| Chinese literature (publication years) | 7% | 50% |
| Pop culture (box office and ratings) | 9% | 50% |
| Astronomy (periods and distances) | 3% | 59% |
Audit rounds
| Round | Passed | Failed | Not tested |
|---|---|---|---|
| 7 Novepoch 30 | 9 | 4 | 5 |
| 4 Novepoch 29 | 7 | 9 | 3 |
| 21 Octepoch 27 · before points | 3 | 3 | 3 |
Audited by approved auditors · evidence on IPFS, hash-checked on-chain.