Checked Nov 11: 27 providers, 4 on watch
The model you paid for.
Every time.
Same model name, different answers. We check every provider and only send your requests to the real thing.
See the evidence or move your cursor over the dots.
Same name. Different answers.
We asked every Kimi K3 provider on OpenRouter the same 106 questions. Most gave Moonshot AI's answer about 65% of the time. Wafer · us gave it 37%. Your bill would look the same either way.
Open the Kimi K3 results →106 questions. Each square is one question. Lit means the same answer as Moonshot AI. Hover a column.
We route around it.
Your requests only go to providers that pass. The rest are skipped until they pass again.
Your appfakemodelcheck
Moonshot AImaker
InferenceNet · fp4passes
InferenceNet · uspasses
Decart · mxfp4passes
Chutes · mxfp4passes
Sail Research · fp4passes
Relace · fp4passes
Sail Research · fp4 uspasses
Phalaoffline
Wafer · uson watch, skipped
- Connect OpenRouter.One click. You keep your key and your bill.
- Change one line.client = OpenAI( base_url="https://api.fakemodelcheck.com/v1", api_key=FMC_KEY, )
- Keep shipping.Same OpenAI API, same models, same bill. No new SDK.
Proof on every response.
Each answer comes back with a receipt: who served it, and what our last check said about them.
- Only providers that pass get your traffic.
- Every result shows its evidence.
- We never store your prompts.
"receipt": {
"model": "deepseek-v4-1-flash",
"provider": "Relace",
"identity": "match",
"checked": "2026-11-11",
"mode": "strict",
"allowed": 10,
"added_ms": 3
}
Every provider. Checked daily.
106 questions per provider, every day. A provider is marked as not matching only after 2 bad days.
| Provider | Same answer as the maker · typical | Result |
|---|---|---|
| DeepSeek | makersame answer twice 98% of the time | Reference |
| Wafer · us | 94%100 of 106 prompts · typical 92% | Matches |
| StreamLake · fp8 | 94%100 of 106 prompts · typical 92% | Matches |
| DeepInfra · fp8 | 93%99 of 106 prompts · typical 92% | Matches |
| DeepInfra · fp8 flex | 92%98 of 106 prompts · typical 92% | Matches |
| Morph · fp8 | 92%97 of 106 prompts · typical 93% | Matches |
| Relace · fp4 | 92%97 of 106 prompts · typical 93% | Matches |
| Particle AI · us | 90%95 of 106 prompts · typical 93% | Matches |
| Particle AI · us flex | 89%94 of 106 prompts · typical 93% | Matches |
| Ionstream · flex | 87%92 of 106 prompts · typical 93% | Matches |
Flat fee. Zero markup.
Tokens stay on your OpenRouter bill.