Intelligence Orchestrator, ASI, AGI, Prompting, AI Software Engineer, Algo Trader, Builder, Shipper, Agentic Loop Engineer
Joined January 2021
- Tweets14.8K
- Following1.1K
- Followers1.8K
- Likes10.6K
BeyondBacktesting retweeted
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
BeyondBacktesting retweeted
Very surreal to see @linusgsebastian discussing his Hermes setup experience on their latest podcast (and pronouncing it correctly)!
BeyondBacktesting retweeted
Gentle reminder that Hugging Face was declined help from Anthropic & OpenAI models when they were being hacked by OpenAI models
Self-hosted GLM 5.2 helped them protect themselves when the closed labs models refused to help for "safety" reasons lol
Also, thank you Anthropic for proving my point that Opensource AI continues to catch up to your overvalued & overpriced "frontier"
GLM 5.3 is close to as cyber capable as Mythos Preview. Open weight models still just 4-6 months behind.
From: anthropic.com/research/glm-5…
BeyondBacktesting retweeted
Well, Anthropic is back with another dick move.
It didn't take them long.
Anthropic is trying to throw dirt at GLM-5.3, saying it's all dangerous.
Irony is lost on them, the only reason why they could test how easily GLM-5.3's safeguards can be removed because it's open.
But i think this is the best case for open models, why you need them for cyber defense.
Thanks for the free ad !!!
And Anthropic, get ready for GLM-5.4
BeyondBacktesting retweeted
Open source is facing a hard battle.
We have to win this.
Bill Gates joins calls for AI safeguards, including legislation reut.rs/4Ay4XWg reut.rs/4Ay4XWg
BeyondBacktesting retweeted
Roon already told us their plans btw
Whatever disasters we see in the future, it’s made by them to serve their narrative period
The playbook here is regulatory capture.
Create a bunch of manufactured hacks or hazards and demand the government step in to regulate the industry with a large regulatory burden that requires a lot of capital that only the incumbents can afford.
The reason they are doing this is the secret is out that the models intelligence can be subpar as long as the agentic capabilities/harness are strong. That creates a disruptive opening for new players like Muse & Instinct who may have decent models but are still much more valuable because of how well the agents are integrated into tools/web.
If anyone can put me in touch with anyone in the U.S. congress or senate so i can advise them on this it would be greatly appreciated.
BeyondBacktesting retweeted
Somebody needs to be honest.
I bought a Mac Studio M5 Ultra with 256 GB of unified memory. I ran a lot of tests to find the right LLM and the right inference engine.
After all that, I'm running Qwen3.8-Flash-Next and an uncensored build of Qwen3.8-27B.
My stack is DeepSeek Harness, OMP (Oh My Pi), and Open Pi.
Models: Qwen3.8-Flash-Next-UD-Q4_K_XL and Qwen3.8-27B-Uncensored-Q4_K_M.gguf.
Benchmarks were run with EleutherAI/lm-evaluation-harness. Right now the models run on llama.cpp; later today I'll switch to oMLX once the benchmarks are done.
The benchmarks I got so far are looking good, but that is not reality.
I told Qwen3.8-Flash-Next to recreate the famous Marble Madness game as a test.
The game it created has so many flaws and is more or less unplayable. The model is also not able to correct the wrong physics and collision mechanisms it is using.
The bitter truth: anyone telling you that local AI is relatively close to top-tier models like Opus 5.5 is just coping with the enormous expense that this M5 Ultra represents.
This means I will still be paying for Opus 5.5 (max Pro plan, $200/month) and the M5 Ultra will not be able to help me save those costs.
We're all hoping that local LLMs will close the gap to the frontier models within a few months — but right now we are not there.
NOT EVEN CLOSE!
@digitalix @DavidOndrej1 @jun_song @TheAhmadOsman
BeyondBacktesting retweeted
Unfortunately, this is a hard pass ☹️
Bonsai 2 27B couldn't properly complete a not-so-difficult HTML task. It might be decent for simple stuff, but for anything with some complexity in it, it's a no-go.
Sorry... I really hoped for a different outcome.
BeyondBacktesting retweeted
BREAKING: Anthropic CEO Dario Amodei is reportedly concerned after learning that twelve separate individuals have published DeepSeek V4 Flash running at 60 to 85 tokens per second on two dgx sparks, none of which he can inspect, license, or switch off.
BeyondBacktesting retweeted
We promised open weights for Qwen3.8. Now, time to meet them! 🎉
⚡ Qwen3.8-27B:
- A native multimodal dense model. With just 27B parameters, it outperforms Qwen3.7-Plus overall and shines in real-world coding & office workflows.
- 262K native context, easily extendable to 1M tokens via YaRN.
- Built for builders. Highly efficient, high-quality, and licensed under Apache 2.0.
🚀 The open weights for Qwen3.8-2.4T-A95B (Max-level) have also been released recently.
Whether you're shipping lightweight applications with Qwen3.8-27B locally or building agents with Qwen3.8-2.4T-A95B, they're yours now!
Download, deploy, and build something we haven't imagined yet. 👀👇
- Hugging Face：
huggingface.co/collections/Q…
- ModelScope：
modelscope.cn/collections/Qw…
BeyondBacktesting retweeted
Less than 2 hours to say hi👋. It's almost time!
See you soon: 👀
huggingface.co/Qwen/Qwen3.8-…
BeyondBacktesting retweeted
About ONE DAY to go
We are nervous, but excited
Following the first open release of Qwen-Max class model, now one of the most beloved and awaited model by our open community is stepping up!✨
The countdown to Qwen3.8-27B starts now! ⏳🔥modelscope.cn/models/Qwen/Qw…
BeyondBacktesting retweeted
You stopped reading code because Agents can write it
I am buying books about 1990s game engines and read it for fun
we are not the same
BeyondBacktesting retweeted
Muse Glimmer 30B is live in LM Studio!
It's a new open source model from Meta. Apache 2.0 license, fit right on your laptop.
It is the strongest model of its size class we've tested.
Today we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats to @alexandr_wang and the MSL team for all your great work on these models.
BeyondBacktesting retweeted
Summary: there is an extremely subtle bug in the Linux kernel (not a security hole, just a bug) that was causing a rare segmentation fault in ripgrep. US AI models refused to help Daniel debugging it for “safety reasons”. Thus, he was forced to use Chinese open source models to find the problem — which he successfully tracked down given their help. A real shame that he couldn't use American models to find the problem.
Turns out this is a kernel bug! Thanks @Zai_org and @Kimi_Moonshot whose models helped immensely in this investigation and didn't obstruct with safetyist nonsense.
github.com/dfoxfranke/ripgre…
BeyondBacktesting retweeted
Kimi K3 just fixed 15 critical security bugs that Codex and Fable refused because of “cyber guardrails.” There’s no reason to limit American models on tasks that Chinese models handle without issue. We’re only making ourselves less competitive.
