Same model, different sellers
Is it the model it says it is?
Many x402 sellers resell the same model name. Asked the same private prompts, sellers serving the same model should answer from the same distribution. These are the names more than one seller the record can buy offers, and how far comparing them has got.
DeepSeek V4.1 Flash4 sellers
| Seller | Model as it names it | Endpoint | Price | Card | Compared |
|---|---|---|---|---|---|
| blockrun | deepseek/deepseek-v4.1-flash |
blockrun.ai/api/v1/chat/completions | $0.002 | UNRATED | not compared |
| telnyx | deepseek-ai/DeepSeek-V4.1-Flash |
x402.telnyx.com/v1/chat/completions/_t/deepseek-v4.1-flash | $0.013259 | UNRATED | not compared |
| singitai | deepseek-v4.1-flash |
ask.singitai.app/v1/chat/deepseek-v4.1-flash | $0.0015 | UNRATED | not compared |
| ai-open-models | deepseek-ai/DeepSeek-V4.1-Flash |
api.ai-open-models.com/v1/chat/completions | $0.002 | UNRATED | not compared |
Not compared yet. Three control runs come first: the same seller a week apart, two resellers of one upstream, and two models known to differ.
not runThe same seller a week apart is not told apart from itself
not runTwo resellers of one upstream are not told apart
heldTwo models known to differ are told apart
GLM 5.3 Flash5 sellers
| Seller | Model as it names it | Endpoint | Price | Card | Compared |
|---|---|---|---|---|---|
| blockrun | zai/glm-5.3-flash |
blockrun.ai/api/v1/chat/completions | $0.002 | UNRATED | not compared |
| telnyx | zai-org/GLM-5.3-Flash |
x402.telnyx.com/v1/chat/completions/_t/glm-5.3-flash | $0.005783 | UNRATED | not compared |
| farouter | glm-5.3-flash |
farouter.tech/v1/models/glm-5.3-flash/chat/completions | $0.002 | UNRATED | not compared |
| laminarity | zai-org/GLM-5.3-Flash |
api.laminarity.uk/v1/chat/completions | $0.001 | UNRATED | not compared |
| ai-open-models | zai-org/GLM-5.3-Flash |
api.ai-open-models.com/v1/chat/completions | $0.002 | UNRATED | not compared |
Not compared yet. Three control runs come first: the same seller a week apart, two resellers of one upstream, and two models known to differ.
not runThe same seller a week apart is not told apart from itself
not runTwo resellers of one upstream are not told apart
heldTwo models known to differ are told apart
How to read this
A passing comparison is weak evidence: a seller can recognise an audit prompt and route it to the real model.
A difference is a measurement, never a verdict. Only attestation proves what ran.
Every call is a paid call in the record, with its transaction and its answer, so the counts can be recomputed.
Sellers are grouped by a person reading each seller's own model list, never by a fuzzy match on names.
Ask the api yourself
curl -s https://api.teppi.xyz/v1/models