Jeremy Janzen

290 posts

Jeremy Janzen banner
Jeremy Janzen

Jeremy Janzen

@Jeremy_Janzen

Helping devs 10x their Claude Code sessions. Founder of https://t.co/oaCBY8pCUg, bringing determinism to AI dev. Sign up for the waitlist today!

Canada Katılım Ocak 2012
75 Takip Edilen115 Takipçiler
Sabitlenmiş Tweet
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
I was constantly fighting Claude Code's context limits. So many /clear and /compacts! So I built a system of subagents. Now my sessions run 10x longer, and the code quality is higher. Here are 6 battle-tested subagents that transformed my workflow (with working code) 🧵👇
English
8
2
40
2.9K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@steipete @Zai_org Just a guess but Anthropic never released their tokenizer so it could be a difference between how Anthropic vs Z count tokens.
English
0
0
2
201
Peter Steinberger 🦞
Peter Steinberger 🦞@steipete·
No clue how they did this tho, the model has 128k token per card, but I'm at 156k already. Either some server trick or they extended the limit. Compactation works fine. No errors. Kudos, @Zai_org !
Peter Steinberger 🦞 tweet media
English
3
1
35
4.1K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@Zai_org I appreciate the honesty posting a benchmark where Claude is better. Albeit looks like a relatively close win.
English
0
0
6
914
Z.ai
Z.ai@Zai_org·
To test models' performance on Claude Code, we ran GLM-4.5 against Claude Sonnet 4 and other open-source models on 52 practical programming tasks. While GLM-4.5 demonstrated strong performance against top open-source models, it secured a 40.4% win rate against Claude Sonnet 4.
Z.ai tweet media
English
5
15
185
32.7K
Z.ai
Z.ai@Zai_org·
Announcing GLM Coding Plan for Claude Code! After seeing the amazing adoption of GLM-4.5 over the past month, we're making it more accessible. Get started: z.ai/subscribe Integration guide: docs.z.ai/scenario-examp… What's new: 1/7th the price of original Claude Code plans 3x more prompts to supercharge your coding workflows
Z.ai tweet media
English
105
152
1.3K
367.4K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@steipete So you’ve had only API failures so far or is it working? I would totally go for $3/month for when I hit Claude limits.
English
1
0
1
325
FluffyCat
FluffyCat@fluffyfatcat1·
@steipete Is it as good as Opus? I tried GPT-5 in Claude Code through OpenRouter and it was abysmal.
English
2
0
1
822
Peter Steinberger 🦞
Peter Steinberger 🦞@steipete·
Been using ~4 parallel agents for codex and don't run into rate limits. Limit seems quite a bit more generous than the Anthropic one. (200/plan)
English
21
2
219
25.1K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@iannuttall For sure. People have built AMAZING products, even solo, without LLMs at all. Or by copy and pasting into ChatGPT 3.5! None of these tools should hold you back if you are dedicated and have a solid idea.
English
0
0
1
198
Ian Nuttall
Ian Nuttall@iannuttall·
you can build something amazing in claude code you can build something amazing in codex you can build something amazing in cursor knowing what to build > the tool you use to build it if it ain’t broke…
jack friks@jackfriks

i switched to claude code a month ago and it one shotted things i was stuck on for hours with on cursor, but how can i know NOW that cursor hasnt gotten better? that it hasnt surpassed claude code in the battle? i could try it myself, since im using claude code on cursor but...

English
22
2
105
14.2K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@GosuCoder Automatic memory is a really hard problem to solve. What memories to save? And when to retrieve them that doesn’t confuse the current context?
English
0
0
0
62
GosuCoder
GosuCoder@GosuCoder·
I’m not sure I like the memory system on ChatGPT. I’ve asked it hypothetical questions about computer upgrades for running LLMs and it keeps thinking I already have my hypothetical setups running LLMs. New chat, for some reason it thinks I already have my RTX 5090 running on my framework desktop. I have my framework desktop and a windows computer with my RTX 5090. Probably worth turning off if you do a lot of hypothetical scenarios.
GosuCoder tweet media
English
6
1
23
2K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@jobergum Should agent benchmarks even be one-shot? Agentic workflows that are real-world usually involve several prompts.
English
1
0
0
52
Jo Kristian Bergum
Jo Kristian Bergum@jobergum·
Are there any one- shot coding agent benchmarks like - codebase - task prompt one shot - solution accepted/rejected
English
3
0
8
1.8K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@alexalbert__ Please have a look at the frequent “Overloaded” errors on Opus 4.1!
English
0
0
11
274
Alex Albert
Alex Albert@alexalbert__·
We just shipped some performance improvements to claude dot ai. Chatting with Claude should start to feel noticeably smoother now⚡️ We're also prioritizing fixing any rough edges or papercuts you run into. Let us know what we should tackle next!
English
202
23
965
132.1K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@vishyfishy2 I don't think its worth it anymore. Claude Code or other agents can walk you through deployment, VPS setup etc.
English
0
0
1
90
f1shy-dev
f1shy-dev@vishyfishy2·
"vercel isn't worth it just get a $5 vps" it's about the DX. who else can deploy a new website, with no prior preparation, in 16 seconds?
f1shy-dev tweet media
English
151
15
749
124.9K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@eliotpaynter @DannyAziz97 I was looking at doing something like this awhile ago but it seemed like an annoyance that the user has to restart their session. But not taking anything away from this - what you built is very cool!
English
1
0
0
20
Eliot Paynter
Eliot Paynter@eliotpaynter·
@Jeremy_Janzen @DannyAziz97 yes, claude —resume <session-id> is how i do it. not sure if there is a work around for that but going to look into it
English
1
0
0
43
Eliot Paynter
Eliot Paynter@eliotpaynter·
Started using claude-prune when @DannyAziz97 shared it - very useful for managing Claude's context limits. Problem: it only kept recent messages. You're deep into a debugging session... Finally! You have a killer plan and are ready to implement it. Shit - context is near full. Standard pruning deletes from the start. Now Claude doesn't know: - What the actual error was - Which fixes you already tried - What the plan even solves So I forked it. Keep beginning + end. Or just the parts with working code. You choose. npm install -g @‌epaynter/claude-prune claude-prune github.com/epaynter/claud…
Eliot Paynter tweet media
English
2
0
7
399
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@steipete @cerebras Thought about that but I still do some Steam gaming so chose Bazzite. I do want to try Omarchy though!
English
0
0
0
61
Peter Steinberger 🦞
Peter Steinberger 🦞@steipete·
Qwen's a good model for refactoring. It's not Claude, but it doesn't need to be, just follow a markdown that Opus wrote.
Peter Steinberger 🦞 tweet media
English
1
0
23
3.8K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@steipete @cerebras Yup it’s for sure hard mode in every way! But CC is actually pretty helpful with that figuring out configs and compiling software.
English
1
0
0
46
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@steipete @cerebras I’ve had some weirdness like that happening in Claude Code since I switched to Linux 🥲
English
1
0
0
81
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@Thomasmarkelly @trq212 For sure you can do it manually. Get Claude to write to todos.md. But Claude Code is about best practices and making it easy for everyone to do that. So having an official solution for this would be nice. (could be optional too if you want to roll your own)
English
1
0
0
49
Thariq
Thariq@trq212·
🫸 STOP COMPACTING, use /clear the biggest difference I see between how the Claude Code team uses CC and the broader community how often they run into compacts instead of keeping long running chats open across multiple tasks, when you finish a task use /clear & start again
English
58
28
466
60.6K
Jeremy Janzen
Jeremy Janzen@Jeremy_Janzen·
@vocamix_hoshino This is really interesting. It seems like its a simulation for now. Are you planning on interfacing it with a LLM?
English
1
0
1
36
Ayomipo
Ayomipo@AyomipoOduntan·
Two major achievements made today: 1. Thanks everyone for helping me reach 30 followers! On track to reach 100 before the end of the year! 2. I reached my first 1000 impression post ever made on this account!
Ayomipo tweet media
English
5
0
10
211

Keşfet

@steipete @Zai_org @fluffyfatcat1 @iannuttall @GosuCoder @jobergum @alexalbert__ @vishyfishy2