←
LEBREL AI36 posts
LEBREL AI avatar

LEBREL AI

@LebrelAI

The AI you can verify. Encrypted. Attested. Uncensored. Signed receipts on Solana. Home of Proof of Edition, an open standard for verifiable inference.

◷ Joined August 2026▧ 29 Photos and videos
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 5

Frontier models. Lower inference costs. Lebrel early access is here, with cheaper model pricing, our own uncensored editions and a signed receipt with every answer. Use Chat or connect through an OpenAI-compatible API: lebrel.ai/

0:08
Qwen avatar
⟳ LEBREL AI retweeted
Qwen@Alibaba_Qwen·Nov 17

Qwen4-Max is now live on @LebrelAI. Come try agentic coding and show us what you build. 😎

The top agentic coding model, live on Lebrel at 7% below Alibaba's list price. Qwen4 Max from @Alibaba_Qwen ranks first on DeepSWE v1.1, ahead of Kimi, GPT and Opus, per Qwen's results. Anonymized, attested, with a signed receipt on each answer. Try it: Lebrel.ai

0:07
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 17

Same 9 models. 17% less overall. Lebrel is cheaper than a leading private-AI provider on 7 of the 9 routed models we both sell, and 17% cheaper on all 9 together: $55.61 vs $67.13 for 1M input + 1M output on each. Plus an attested router and a signed receipt on every answer.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 16

The top agentic coding model, live on Lebrel at 7% below Alibaba's list price. Qwen4 Max from @Alibaba_Qwen ranks first on DeepSWE v1.1, ahead of Kimi, GPT and Opus, per Qwen's results. Anonymized, attested, with a signed receipt on each answer. Try it: Lebrel.ai

0:07
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 13

One key, one balance, every model. Try the uncensored edition from $0.13 in / $0.53 out per million tokens, free daily 15:00 to 19:00 UTC during the open beta, or any routed model at its listed rate. Every answer comes with a signed receipt: lebrel.ai.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 13

Not every job needs an uncensored model. The same key reaches 8 models from DeepSeek, Moonshot, Alibaba and Z.ai, including Kimi K3.1 Pro, GLM-5.3 and Qwen3.8 Max, through our attested router: the lab sees the request, not who sent it.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 13

The model underneath leads on agent work. In DeepSeek's published table, V4.1 Flash 1017 leads Opus 5, GPT-5.6 Sol, GLM-5.3 and K3 on DeepSWE (77.0), AutomationBench (60.5), Agents' Last Exam (35.4) and Apex Bench (72.1). Scores published by DeepSeek, not measured by Lebrel.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 13

Frontier capability shouldn't stop at a refusal. Lebrel serves an uncensored edition of DeepSeek V4.1 Flash, with no refusal layer added on top. In Chat and the API. Free every day 15:00 to 19:00 UTC during the open beta.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 12

AI you can afford to keep using. From a conversation in Chat to the model powering your app, Lebrel keeps inference costs down and gives you access to our own uncensored editions, without an added refusal layer. Get started: lebrel.ai

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 9

Start your Lebrel collection. Try our own uncensored editions free every day from 3-7pm UTC, in Chat and through the API. Unlock badges as you use the models, return for another session and share your experience. See what you can earn: quests.lebrel.ai

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 6

Four hours of free AI, every day. Use Lebrel’s own uncensored model editions in Chat and through the API, free everyday from 3-7pm UTC during this early access phase. Explore missions and collect badges as you use Lebrel: quests.lebrel.ai

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 5

Try Lebrel today. Our own uncensored editions are free to use in Chat and through the API, every day from 3-7pm UTC. Start your First Run, explore the missions and collect badges: quests.lebrel.ai

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 5

Already building with an OpenAI-compatible API? Set your base URL to api.lebrel.ai/v1 and use your Lebrel API key. Bring Lebrel’s models into your app or agent through a familiar interface. Lower inference costs without rebuilding your integration.

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 5

Lower prices, with a record behind each answer. Every response comes with a signed receipt linking it to a model or route manifest. Hourly checks monitor model routes for changes or mismatches. Check a receipt: lebrel.ai/verify

Post attachment
LEBREL AI avatar
LEBREL AI@LebrelAI·Nov 5

Lebrel also runs its own uncensored editions of DeepSeek V4.1 Flash, Qwen 3.8 27B and GLM-5.3 Flash. These editions have no added refusal layer, with input encrypted from your device to the model. Available in Chat and through the API.

Post attachment