Professional larper @Brave @Obscuravpn
Pluto
Joined May 2023
- Tweets17.6K
- Following357
- Followers21.5K
- Likes125K
Keith retweeted
Its my favorite LLM at the moment, Kimi is really positive, but doesn't feel agreeable
So many LLMs just agree with you on everything even though you are completely wrong about, I haven't had that experience with Kimi K2, overall its a very good model ^^
Some people say its even better at coding than Sonnet 4
Keith retweeted
Not a vuln in QUIC but in HTTP/3 QPACK.
@obscuravpn is not affected.
We decided to stick with just QUIC to minimize complexity + attack surface. That decision paid off today.
Sorry everyone... We're announcing a 0-Day vuln in QUIC that allows anyone to remotely turn off a server running these libraries. I swear we didn't plan this.
foxio.io/blog/header-accordi…
Keith retweeted
Can I sponsor you? I’m doing a Keithcon next month about how KeithAI will be what KeithBrowser failed to Keith. Keithing so, KeithAI would Keith when Keith in Keith against Keith Keith Keith. Please, KeithAI Keith Keith Keith…keight kekekeke, keith keithing keithed.
I would email @JensenHuang personally but I don't have motion like that unfortunately so I just reached out to the company itself.
Does anyone use Venice? If so could someone tell me the usage you get?
DeepSeek V4.1 Flash just got a serious upgrade.
@deepseek_ai’s 1017 outscores leading open-source and frontier models on nearly every benchmark.
Same model in the Venice app, simply upgraded and available via the API as deepseek-v4-1-flash-1017.
Fully private.
Keith retweeted
Aloha!🌺 Meet Ornith-1.5-122B: No Tradeoff Between Strong and Fast
Stronger. Faster. Leaner. Ornith-1.5-122B delivers frontier-level capabilities with exceptional inference efficiency. With superior instruction following, stronger task decomposition, and deeper understanding of real-world workflows, it achieves state-of-the-art performance among open-source models of comparable size, while running on a single NVIDIA DGX card!!
Open weights under the MIT license.
Coding
✅SWE-bench Verified: 0.845
✅TB2.1: 0.813
✅CyberGym: 0.800
✅NL2Repo: 0.512
Agentic
✅HLE: 0.446
✅BrowseComp: 0.785
✅MCP-Atlas: 0.746
✅Toolathlon: 0.708
🤗Hugging Face: huggingface.co/ornith-ai/Orn…
We can’t wait to see what you build with Ornith-1.5-122B.
To anyone who reaches out to me for sponsorships, the answer 90% of the time is going to be no.
I don't want your money (yes I do) but I won't take it.
If you make something genuinely cool and awesome that I know people would love, I'm willing to hear you out.
Thanks.
I will say if you want to sponsor a project that enables me to work more or I can use your product to accelerate work I'm doing, I'm much more likely to shout you out.
Especially if you are a hardware provider.
I don't want to be the only one who wins during a sponsorship, if my community doesn't also win.
I've lost and sold out.
Keith retweeted
Grateful to @deepseek_ai for releasing DeepInfer, initially built primarily on @vllm_project, under Apache 2.0 and crediting the open-source community!
We've worked hard on DeepSeek V4.1 Flash inference efficiency and will keep improving with ideas from DeepInfer! Stay tuned 🙌
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
I genuinely do believe that every model should internally have a True or False state inside of them to reduce hallucinations drastically, obviously there are ways to do this externally but it being built into the model would be cool.
I have a general idea on how I would want something like that to look like architecturally and might make small model showcasing to showcase it.
MiniMax-M3.1-Flash scores 42 on the Artificial Analysis Intelligence Index, 13 points higher than MiniMax-M3 and less than half the cost per task. It scores level with GLM-5.3-Flash but behind DeepSeek V4.1 Flash 1017
MiniMax-M3.1-Flash is @MiniMax_AI's new Flash reasoning model, which we benchmarked at its max effort setting. It scores 42 on the Artificial Analysis Intelligence Index, level with GLM-5.3-Flash (max, 42), just ahead of Gemini 3.8 Flash (high, 41), and behind DeepSeek V4.1 Flash 1017 (max, 45)
Key takeaways:
➤ Complex reasoning improved substantially compared to M3: MiniMax-M3.1-Flash scores 48% on Humanity's Last Exam, ahead of GLM-5.3-Flash (max, 40%) and DeepSeek V4.1 Flash 1017 (max, 43%), and level with Gemini 3.8 Flash (high, 48%). On CritPt it scores 22%, ahead of GLM-5.3-Flash (15%) and Gemini 3.8 Flash (18%) and just behind DeepSeek V4.1 Flash 1017 (24%). These are up from MiniMax-M3 which scored 39% and 4% respectively
➤ Fewer output tokens, lower cost per task: MiniMax-M3.1-Flash used ~189M output tokens to run the Intelligence Index, considerably fewer than GLM-5.3-Flash (max, ~324M), DeepSeek V4.1 Flash 1017 (max, ~224M) and Gemini 3.8 Flash (high, ~334M). Combined with pricing of $0.15/$0.60 per 1M input/output tokens, this brings its cost to ~$0.22 per Intelligence Index task, vs ~$0.28 for GLM-5.3-Flash, ~$0.26 for DeepSeek V4.1 Flash 1017 and ~$1.24 for Gemini 3.8 Flash, and less than half MiniMax-M3's ~$0.51
➤ High AA-Omniscience accuracy, offset by a high hallucination rate: MiniMax-M3.1-Flash answers 56% of questions correctly on AA-Omniscience, our benchmark measuring factual recall and hallucination, vs 28% for GLM-5.3-Flash (max) and 43% for DeepSeek V4.1 Flash 1017 (max). It attempts almost every question, resulting in a hallucination rate of 93% and an overall score of 16 on AA-Omniscience
Key model details:
➤ Context window: 1M tokens
➤ Multimodality: Text and image input, text output
➤ Pricing: $0.15/$0.60 per 1M input/output tokens, with cached input at $0.015/M
➤ Availability: MiniMax first-party API, with MiniMax announcing an open weights release on October 23rd
➤ Licensing: Closed weights currently, with MiniMax announcing a weights release on October 23rd
