Joined April 2018
- Tweets 17,356
- Following 422
- Followers 2,643
- Likes 32,631
3,899 Photos and videos
Most Claude Code demos skip the part that actually matters.
Claude Code dynamic workflows can spin up a lot of agents fast.
Would you run this on a real repo?
2
3
57
Jun 13
Now that Fable/Mythos is banned, can everyone else launch new models without showcasing benchmarks?
That would be great, most of them are BS anyways, we already know the model will be better.
2
39
Jun 13
AI agents do not need more hype. They need a workflow that fails safely.
AI coding agents need workspaces, panes, browser surfaces, and notifications instead of random terminal tabs.
Would you run this on a real repo?
2
1
4
113
CodeGlitch retweeted
Jun 13
Funny, that was actually the first model from Anthropic that didn't suck
Wonder what this will mean for OpenAI next big release
Back to Codex?
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…
1
3
104
Jun 9
Jun 9
Claude launches Fable 5 a "Mythos class model", yet cant provide enough compute or api rate limits so that you can use their products with more than a couple of sessions running.
GG
1
48
Jun 13
Jun 13
Funny, that was actually the first model from Anthropic that didn't suck
Wonder what this will mean for OpenAI next big release
Back to Codex?
22
Jun 12
Context Beats The Model
The real AI operating system upgrade is persistent context, not a newer Claude model.
Would you run this on a real repo?
21
Jun 12
Most Claude Code demos skip the part that actually matters.
Claude Code dynamic workflows can fan work out across parallel agents, but the useful question is whether the orchestration is actually good enough.
Would you run this on a real repo?
2
47
Jun 12
Stop Fine-Tuning Local AI Wrong
Fine-tuning local AI is useful, but the checklist matters more than the hype.
Would you use this, or skip it?
1
1
42
Jun 11
Do Not Give Opus Every Task
Opus 4.8 looks stronger, but expensive models should be reserved for architecture, hard bugs, and review.
Would you use this, or skip it?
1
2
49
Jun 11
AI agents do not need more hype. They need a workflow that fails safely.
Would you run this on a real repo?
2
1
48
Jun 11
AI agents do not need more hype. They need a workflow that fails safely.
DeepSWE matters because it tests AI coding agents on handwritten repo tasks instead of rewarding memorized public GitHub patches.
Would you run this on a real repo?
1
1
75
Jun 10
Most Claude Code demos skip the part that actually matters.
Claude Code looked stronger on front-end polish, but that does not mean it wins every coding-agent task.
Would you run this on a real repo?
1
2
54