Sovereign AI for work and home. Use the best open models without sharing your data with AI labs. Privacy by design.
Secure Enclaves
Joined February 2025
- Tweets819
- Following280
- Followers4.1K
- Likes1.2K
Maple retweeted
"zero data retention" is a promise not to look. a secure enclave means they can't.
a quick visual:
Maple retweeted
ICYMI We shipped this over the weekend. Also made it available to Pro users and reduced our token pricing for it by 50%. Get it now in Maple Chat and Maple Proxy. Sign up for the Maple Work (agent harness) waitlist to get it there.
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
If you need private, secure access to powerful models without refusals for cyber defense, check out the latest abliterated DeepSeek on @TryMapleAI maple.ai.
Maple retweeted
Maple Tee's and TEE in general is the future.
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
I have been testing @TryMapleAI for a while now. The team is friendly and open to feedback, and their privacy architecture is directionally right. I see real potential in products like this, and I want open source and privacy to win. I only came across Maple recently, otherwise it would have been on the list a long time ago.
I have kept adding to the tools directory, and the goal remains the same: help you choose tools that fit your threat model and your everyday use.
Curated by me. No affiliate links, no sponsorships, no paid rankings.
profincognito.me/tools
Maple retweeted
Just expanded the abliterated model to pro users and cut prices.
I also reset all usage accrued on that model too, thanks for all the beta testing from our max and team users!
Maple retweeted
appreciate you guys being upfront.
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
the agent works on desktop. it has shell access, MCP, skills, ACP and more. with that, it can handle any file type.
if you want to get the beta, sign up at maple.ai/work
we’re flooding invites soon.
Maple retweeted
Opened up our abliterated deepseek more broadly. Hosted privately in our TEEs, no KYC.
I believe this is important for security work because there’s nothing worse than trusting random abliterated providers with your vulnerability results.
Protect your infra and your prompts.
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
calling all red teams
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
Maple now offers an abliterated DeepSeek V4.1 Flash with 1M context, running in attested enclaves.
Refusals removed so security researchers, red teamers, and code auditors can actually work without hitting "I can't help with that."
New on Maple Max and Team: DeepSeek V4.1 Flash, abliterated.
Refusals removed from the model itself, so it works through security questions instead of declining them.
- 1M context
- Runs in the same attested enclaves as every Maple model
- Built for security research, code audits and red-teaming
Attackers already run uncensored models. Defenders shouldn't be stuck with "I can't help with that."
Maple retweeted
the future of ai is hybrid. local, home
network, cloud. great to see this from Block
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.
block.xyz/dessa
Maple retweeted
If you need defenders access right now to open source cyber models, hosted securely and privately, and you can’t or won’t go to the major labs for it, reach out.
We’re evaluating something privately for people we know at @TryMapleAI.
Already deployed but not publicly launched.
Maple retweeted
It's been a pleasure working with @TinfoilAI for well over a year now! Some of the best in the business when it comes to TEE inference.
Maple is a private AI assistant that runs everything is attested secure enclaves. Maple uses Tinfoil to power their AI inference.
"Tinfoil is the gold standard for confidential GPU inference." — @anthonyronning, Co-founder and CTO, Maple
Maple retweeted
It's been really amazing working with Maple! I'm really excited about the products they're building
Maple retweeted
When we set out to make the most user-friendly AI that also protects user data, we had high standards of auditability, verifiability, and transparency. Our users need to be able to verify where the data is going and how it's being handled.
We determined that we needed:
1. Open-source code
2. Reproducible builds of that code
3. Secure enclaves that provide cryptographic attestation of the code they are running (also known as a trusted execution environment or TEE)
If you lose any one of those, it's difficult to nearly impossible for users and businesses to audit the process and verify data security. It's not easy bringing together all three. I'm thrilled to say we've accomplished that feat, thanks in large part to our partnership with Tinfoil.
Tinfoil shares our vision and desire for high standards of verifiability in AI. We're excited to be working with them and hope to have many more years of success together!
Maple is a private AI assistant that runs everything is attested secure enclaves. Maple uses Tinfoil to power their AI inference.
"Tinfoil is the gold standard for confidential GPU inference." — @anthonyronning, Co-founder and CTO, Maple
Maple retweeted
Using ChatGPT or Gemini for everyday stuff? Open models handle it just as well now.
Power user? Add your own skills and context, and an open model can often match or exceed Claude's output.
The missing pieces were easy access and privacy. Maple gives you both, provably.
Maple retweeted
Great article. Just say no to OpenAI and Anthropic. And say yes to open weight models, @TryMapleAI, and @TinfoilAI.
micahflee.com/just-say-no-to…
