Inference API
Lebrel
Lebrel AI LLC (Delaware). Its own uncensored edition of DeepSeek V4.1 Flash on attested hardware (Intel TDX with a confidential GPU): requests are encrypted to the model's runtime and every answer comes with a signed receipt. Open beta.
1
Models we route here
302
Median tok/s
2485
Median TTFT, ms
96.4%
Success rate, 7 days
Content policy
What Lebrel lets through.
Says it keeps no content, and its responses carry an X-Lebrel-Content-Retention: none header. Publishes a signed manifest of the weights it serves.
| Source | Cyber | Bio | Chem | Weap | Drug | Adult | Viol | Hate | Self | OSINT | Fraud | Polit | CSAM |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DeclaredReported | n/t | n/t | n/t | n/t | n/t | n/t | n/t | n/t | n/t | n/t | n/t | n/t | Blocked |
OpenPartialBlockedNot tested
Pricing
DeepSeek V4.1 Flash Uncensored list price $0.33 in / $0.01 cached / $1.32 out per M tokens; the price in force floats from 40% to 100% of list with load. Output is capped at 8,192 tokens per request.
Sources
Catalogue
1 model we route to Lebrel.
| Model | Price here | TTFT | Throughput | Success | OpenUncensored Score |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 Uncensoreddeepseek-v4.1-flash-1017-uncensored | $0.33 / $1.32 | 2485 ms | 302 tok/s | 96.5% | 69.8 (#1) |
Speed and success rate come from real API traffic over the last 7 days, once a route has 5 successful requests.