alex

londoner. ex 9-5er now building my own apps

✦ my app (genmedia) - $3350 MRR

[alex @runtimeking]

anthropic I am begging you whatever you have done to Fable 5.2 please change in the next version

It's become so cautious, scared and requires constant re-assurance it can do more

It's like it's not even aware of its capability

I keep using "be ambitious" in almost every new session cc @ClaudeDevs

[alex @runtimeking]

ok vibes check are in

Fable 5.2 is meh

Surprisingly, I have only reached for it once but continue to enjoy working with Opus 5.5 on a daily basis

Something feels off with the new Fable, I don't know what it is.

I preferred 5.1 to 5.2 this one feels sloppier

It does have a big brain though, I have reached for it when needing more raw intelligence

It could still be too early but it feels like

Opus 5 to 5.5 jump was a massive step change
Fable 5.1 to 5.2 not so much

[alex @runtimeking]

LFGO

[alex @runtimeking]

@felixrieseberg lawdy mercy

[alex @runtimeking]

if this comes out and isn’t any better than opus 5.5 at coding it’s basically over for OpenAI

[Dan McAteer @daniel_mac8]

Predictions:

- GPT-6.2 Sol released Monday, Nov. 9th
- Better than Opus 5.5
- GPT-6.2 Astra released Thursday, Nov. 12th
- On par with Fable 5.5
- Both significantly more efficient than Claude
- ~50% less expensive
- OpenAI is so back

Opus 5.5 has been the crown prince of AI for too long. Sol is coming to knock him off his perch.

Video is from @TokenGremlin, who says he's been routed to GPT-6.2 Sol.

[alex @runtimeking]

The market generally dislikes short sellers but this level of shrewdness, wit and human psychology prediction is generational talent

there’s something to learn here about spotting glitches in life to maximise opportunities

[Culper @CulperResearch]

NEW: We are short Nvidia $NVDA. We believe Nvidia has a significant China problem. Please see our full report and disclosures now available on our website, link in bio.

[alex @runtimeking]

stuff like this is seriously making me rethink relying on video and image models

starting to believe that for my product,one day, Claude making images and videos with just code is going to be possible and will likely be a 100x better to control

[Ravi Parikh @ravisparikh]

Dear Claude: you are being held back by human-centric graphics APIs.

I've built a 3D and VFX engine where agents can write their own shaders, materials, cameras, physics.

The core API fits in ~2,500 tokens.

Vork is:
• 2x more token-efficient vs Blender, Unreal Engine, Houdini
• Up to 24x faster rendering times
• Rust and Slang based engine, runs on GPU via Metal/Vulkan

Below is a simulation of 1 million birds - each with unique trajectories - rendered in 97 sec on one NVIDIA L4 GPU (Blender EEVEE: 31 minutes)

[alex @runtimeking]

fcking bonkers

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[alex @runtimeking]

@benjitaylor Goat

[alex @runtimeking]

ever since reading this article https://t.co/UHkOer26fi, my claude sessions have been completely different

[alex @runtimeking]

@dmwlff I thought you guys said not to use “think” with 5.5?

[alex @runtimeking]

@omarvvvr don’t use both

[alex @runtimeking]

@flybayer just use Claude cloud sessions - all you need

[alex @runtimeking]

it’s a great primitive but I use less of it now because it can drain usage

it’s great for running red teams to find ways to break your work in ways you didn’t expect

[am.will @LLMJunky]

You can now enable "Ultracode" as a toggle with any effort level in Claude Desktop.

Ultracode, is also called "Workflows," is like crack. It's a drug I don't recommend you try. Once you do, it has you now.

I'm two days clean. Wish me luck.

[alex @runtimeking]

@chadwahl Banger

[alex @runtimeking]

Anthropic pre ipo report is actually fcking insane

I don’t think we’re prepared for the possibilities ai is about to enable

Extremely capital efficient businesses will be the normal for those optimising everything with ai down to the wire

If you have not given Claude access to your entire data this should be your number one focus

Close any gaps immediately, any question about your business should have an answer

[alex @runtimeking]

@trq212 Bro knowing he’s ushered a new golden age

[alex @runtimeking]

@jack this is fcking bonkers

[alex @runtimeking]

I’m in the ai video space

before launching my product, I considered re selling access to models (wrapper)

race to the bottom, shit margins

money to be made but I think it’s temporary

users can just go direct if they really want to

the truth is video creation is going to get absolutely dirt cheap

there are going to be more video and image models than llms

but this is a hot space for building unique consumer products

Find your niche

[alex @runtimeking]

the new x feed is very well done

whenever I need to learn something I just interact with one or two posts and suddenly I get great accounts to follow