Model checks
GLM 5.3, provider by provider. Model →
last audit 18 Nov · epoch 31 · evidence checked on-chain26 of 27
providers answered as the model
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.
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
26
SAME
Failed
1
DIFF
Not tested
4
the audit could not run
By provider
Compared with z-ai/glm-5.3 on OpenRouter · pinned to z-ai/fp8 · reasoning minimal
| Provider | Differed↑ | × noise | Fail line | Noise | Proof |
|---|---|---|---|---|---|
| InspectantSAME | 6% | 0.5× | 25 under | 11.3% | ↗ |
| 0G ConfidentialSAME | 6% | 0.5× | 25 under | 11.3% | ↗ |
| Super SeederSAME | 7% | 0.6× | 22 under | 11.3% | ↗ |
| bartly.eth64.deSAME | 8% | 0.7× | 24 under | 11.3% | ↗ |
| Meridian InferenceSAME | 8% | 0.7× | 22 under | 11.3% | ↗ |
| klinnSAME | 9% | 0.8× | 20 under | 11.3% | ↗ |
| surplus-hiveSAME | 9% | 0.8× | 19 under | 11.3% | ↗ |
| AIXSAME | 9% | 0.8× | 26 under | 11.3% | ↗ |
| Unified Rack · DeepSeek 240 tok/sSAME | 10% | 0.9× | 28 under | 11.3% | ↗ |
| SCloudSAME | 11% | 0.9× | 27 under | 11.3% | ↗ |
| LLMartSAME | 11% | 0.9× | 27 under | 11.3% | ↗ |
| ZLKPro-ApiSAME | 12% | 1.1× | 25 under | 11.3% | ↗ |
| Night Harbor AISAME | 12% | 1.1× | 25 under | 11.3% | ↗ |
| Open ForgeSAME | 12% | 1.1× | 25 under | 11.3% | ↗ |
| engySAME | 12% | 1.1× | 25 under | 11.3% | ↗ |
| RS AISAME | 12% | 1.1× | 20 under | 11.3% | ↗ |
| SapiensSAME | 12% | 1.1× | 21 under | 11.3% | ↗ |
| api.contactboxtools.meSAME | 13% | 1.1× | 24 under | 11.3% | ↗ |
| ▲ Apex AntSAME | 13% | 1.1× | 19 under | 11.3% | ↗ |
| Hana Gateway ✅SAME | 14% | 1.2× | 16 under | 11.3% | ↗ |
| PearlSAME | 17% | 1.5× | 17 under | 11.3% | ↗ |
| Relay WorksSAME | 17% | 1.5× | 17 under | 11.3% | ↗ |
| Vikey SellerSAME | 18% | 1.6× | 16 under | 11.3% | ↗ |
| BestSAME | 18% | 1.6× | 15 under | 11.3% | ↗ |
| PhalaSAME | 18% | 1.6× | 14 under | 11.3% | ↗ |
| RedPillSAME | 20% | 1.8× | 12 under | 11.3% | ↗ |
| dealerDIFF | 39% | 3.4× | +18 over | 11.3% | ↗ |
| attacompute.ai not tested · timed out | — | — | — | — | ↗ |
| incharge-austin not tested · provider unreachable | — | — | — | — | ↗ |
| Bosphorus Compute not tested · rate limited | — | — | — | — | ↗ |
| zro not tested · provider unreachable | — | — | — | — | ↗ |
9 older checks · 3 failed · not counting now
| Provider | Differed↑ | × noise | Fail line | Noise | Proof |
|---|---|---|---|---|---|
| 0G ConfidentialSAME 7 Nov · expired | 6% | 0.8× | 17 under | 8.5% | ↗ |
| Fire Ant 🔥🐜SAME 7 Nov · expired | 7% | 0.8× | 21 under | 8.5% | ↗ |
| Vector StudiosSAME 7 Nov · expired · auditors split: 1 DIFF, 1 SAME | 11% | 0.9× | 25 under | 11.5% | ↗ |
| 💬 BanterSAME 4 Nov · expired | 13% | 1.4× | 11 under | 9.4% | ↗ |
| JOI SEEDSAME 7 Nov · expired | 13% | 1.5× | 16 under | 8.5% | ↗ |
| Edith AIDIFF 7 Nov · expired | 45% | 5.3× | +25 over | 8.5% | ↗ |
| TokenSaleDIFF 7 Nov · expired | 55% | 4.7× | +32 over | 11.5% | ↗ |
| antseed-cosmic-gecko-c46dDIFF 7 Nov · expired | 58% | 5.0× | +36 over | 11.5% | ↗ |
| antseed-aggregator 4 Nov · expired · not tested · provider unreachable | — | — | — | — | ↗ |
Where providers miss
| Subject | Passed providers missed | Failed providers missed |
|---|---|---|
| Programming (release years) | 21% | 31% |
| Cryptography (parameters) | 12% | 30% |
| Chinese geography (measurements) | 8% | 17% |
| Astronomy (periods and distances) | 17% | 38% |
| Chinese literature (publication years) | 9% | 0% |
| Physics (material properties) | 11% | 35% |
| Pop culture (box office and ratings) | 12% | 50% |
| Biology (chromosome numbers) | 17% | 50% |
| Internet culture (dates and numbers) | 8% | 23% |
| Medical (drug half-lives) | 3% | 63% |
Audit rounds
| Round | Passed | Failed | Not tested |
|---|---|---|---|
| 17 Novepoch 31 | 26 | 1 | 4 |
| 7 Novepoch 30 | 22 | 3 | 5 |
| 4 Novepoch 29 | 22 | 0 | 9 |
Audited by approved auditors · evidence on IPFS, hash-checked on-chain.