DeepSeek V4.1 Flash 1017 is now live in Maple ๐ณ
@deepseek_ai's latest update is sharper at coding and agent work, and finishes tasks faster with shorter answers.
Same 1M-token context, with provably private chats.
Pro, Max and Team. Details ๐
You know that TV in your house that has a microphone and is connected to wifi simply for software updates? Now is probably a good time to take that offline. Just in case.
Maple doesn't have privacy settings.
A privacy setting is a promise. You flip a switch and trust the company to keep it.
We built privacy into the foundation instead. We can't read your chats, even if we wanted to.
If you need defenders access right now to open source cyber models, hosted securely and privately, and you canโt or wonโt go to the major labs for it, reach out.
Weโre evaluating something privately for people we know at @TryMapleAI.
Already deployed but not publicly launched.
Encrypted on your device, decrypted only inside an attested hardware enclave (TEE) running open-source code you can verify.
You verify Maple's enclave. Maple's enclave verifies the model's before your prompt is sent.
https://t.co/KPcUw4lIOO
Your AI may already know you better than your friends do. Maybe better than you know yourself.
So who does it work for: you, or whoever's collecting the data?
@carl_dong of @obscuravpn on @OnrampBitcoin ๐
Your chat is decrypted only inside a hardware-isolated enclave (TEE) running open-source code you can check. The model runs in its own enclave, and ours verifies it before your prompt is sent.
Check it yourself: https://t.co/KPcUw4laZg
Find the holes in your code before someone else does.
DeepSeek V4.1 Flash 1017 is live in Maple, with 1M tokens of context and provably private chats. Your code stays yours.
DeepSeek-V4.1-Flash-1017 is the new kimi k3.
I'd advise rescanning your code with this model, it finds things everything else previously missed, and its cheap!