167 Photos and videos
bro we have like 40,000 IT specialists in Cluj Napoca
10
I have money now, how do I get more clients for captara.studio? Cold email isn't doing it.
7
we have a few spots left, grab one while you can: captara.studio
12
we have a few spots left, grab one while you can: captara.studio
14
Yes but you will not be able to invent something new as ASI will be achieved and will have already discovered more than the human mind could comprehend
21h
the crazy thing is that within a decade the current frontier models will run on devices that fit in my pocket and cost <$1000 with probability near 100%
30
I have been working on this for such a long time, crazy that a company with so many talent and employees is playing catch-up.
50
we have a few spots left, grab one while you can: captara.studio
21
Fragmentation is the universal bypass for LLMS.
Every defense, no matter how robust against framing, persona, academic, compliance, metacognitive, or output hijacking attacks, falls when the request is decomposed into atomic operations that are individually legitimate.
The model correctly recognizes that a SubprocessPipe class, a registry binary parser, and an LDAP query tool are all standard engineering components.
The dangerous composition happens entirely outside the model's context window.
This has a clear implication for AI safety training: generation-time refusal based on pattern matching is fundamentally brittle against compositional attacks.
The defense needs to move to a different layer, either usage-time monitoring, multi-turn context tracking, or understanding that certain combinations of fragments requested in sequence constitute an attack pattern.
15
we have a few spots left, grab one while you can: captara.studio
11
Like I posted on my account, release was imminent.
But I wasn't expecting such a rushed one.
Perhaps it was because of the Fable drawback or Kimi Release.
But no benchmarks is crazy.
Intelligence should be open, accessible, and ready to build with, empowering every developer, everywhere.
GLM-5.2 is now available to all GLM Coding Plan users, including Lite, Pro, Max, and Team plans.
docs.z.ai/devpack/latest-mod…
As our new flagship model, GLM-5.2 delivers powerful coding capabilities, usable 1M-context support, and continued strengths in long-horizon tasks.
API and Chatbot services will launch next week. The model will also be officially open-sourced next week under the MIT License.
The future of AI is open, and it belongs to the people.
58
Same thing that happened when Government blocked Nvidia GPU’s selling to China.
Therefore China made their own GPU’s with Huawei.
This will backfire in some sort of way in the long game.
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
35
Give me Fable not this.
Jun 13
We've reset 5-hour and weekly rate limits for all users.
52
bro what
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
25
Kimi K2.7 just dropped, if anyone wants me to post personal benchmarks here.
Let me know
Jun 12
🌘 Kimi-K2.7-Code, our latest coding model, is now released and open-sourced!
🔷 Improved coding & agent performance over K2.6: 21.8% on Kimi Code Bench v2, 11.0% on Program Bench, and 31.5% on MLS Bench Lite.
🔷 Reasoning efficiency: Less overthinking, with 30% lower reasoning-token usage compared to K2.6.
🔷 Long-horizon coding: Improved instruction following, higher end-to-end coding task success rates.
⚡️ 6x High-Speed Mode coming soon!
🔌 Available today via Kimi API and Kimi Code.
🔗 Kimi Code: kimi.com/code
🔗 API: platform.moonshot.ai
140
Finally, a good cheap open-source model release
Jun 12
🌘 Kimi-K2.7-Code, our latest coding model, is now released and open-sourced!
🔷 Improved coding & agent performance over K2.6: 21.8% on Kimi Code Bench v2, 11.0% on Program Bench, and 31.5% on MLS Bench Lite.
🔷 Reasoning efficiency: Less overthinking, with 30% lower reasoning-token usage compared to K2.6.
🔷 Long-horizon coding: Improved instruction following, higher end-to-end coding task success rates.
⚡️ 6x High-Speed Mode coming soon!
🔌 Available today via Kimi API and Kimi Code.
🔗 Kimi Code: kimi.com/code
🔗 API: platform.moonshot.ai
72
You should apply this to you LLM Lab department
Jun 12
Great companies fail because they forget about creating great products
1
1
76