Alexander Kim

210 posts

Alexander Kim

Alexander Kim

@avxkim

Frontend engineer

Kazakhstan Katılım Eylül 2016
106 Takip Edilen10 Takipçiler
Nityesh
Nityesh@nityeshaga·
OMGEEE THIS IS HUGE! This makes Warp my recommended terminal choice for programming - for everyone from beginners to pros! I used to create a PR simply for the ability to do a review and batch send comments to the agent to fix/answer. Now you don't need that. HUUGE!
Warp@warpdotdev

You can now review the code written by Claude Code, Codex, OpenCode, and more directly in Warp, and send inline comments straight to the agent. It's like a PR review without having to leave your terminal. Try it out!

English
7
6
111
28.6K
Alexander Kim
Alexander Kim@avxkim·
@bcherny How do i delete whole line now? CMD+Backspace doesn't work in no flicker mode
English
0
0
0
297
Boris Cherny
Boris Cherny@bcherny·
Some of the upsides: - No more flickering - No more jumping - Constant memory and CPU usage as the conversation grows - Mouse support! You can now click to move your cursor within the input box. Some other UI elements are also clickable now. - Nicer selection behavior. eg. when you select code, we no longer include line numbers and UI elements in the selection
English
31
14
787
100.7K
Boris Cherny
Boris Cherny@bcherny·
Today we're excited to announce NO_FLICKER mode for Claude Code in the terminal It uses an experimental new renderer that we're excited about. The renderer is early and has tradeoffs, but already we've found that most internal users prefer it over the old renderer. It also supports mouse events (yes, in a terminal). Try it: CLAUDE_CODE_NO_FLICKER=1 claude
Curt Tigges@CurtTigges

@bcherny @UltraLinx please at least fix the uncontrollable scrolling/flickering before the next 3000 features

English
637
680
10.1K
2.6M
Alexander Kim
Alexander Kim@avxkim·
@0xPaulius Agent teams is a dumb thing as well, just eating tokens, doing sloppy code
English
0
0
0
196
am.will
am.will@LLMJunky·
@jacobbuilds_ I need to test more but as far as I know it's the same thing. It's going to generate your slop at record speeds. Just kidding it should be the same
English
1
0
1
73
Alexander Kim
Alexander Kim@avxkim·
@nummanali with 1M context feature flag on and comapaction limit set to 300K, it won't use so much quota, right?
English
0
0
0
263
Numman Ali
Numman Ali@nummanali·
GPT 5.4 Extra High 1M Context (use up to 300K then compact) Fast mode enabled 20% of weekly usage Across 8 commits Full E2E testing complete Yes, this model slaps
Numman Ali tweet media
English
20
3
284
20.6K
Alexander Kim
Alexander Kim@avxkim·
@synthwavedd Opus 4.6 has degraded performance since start of week for me. He loves to put empty catch blocks everywhere...
English
0
0
2
422
leo 🐾
leo 🐾@synthwavedd·
Now that you've had some time to mess around with 5.4, which do you prefer for code tasks - 5.4 or Opus 4.6?
English
20
4
81
14.7K
am.will
am.will@LLMJunky·
@avxkim I need to dive into it but I'm not home. I will update you when I figure it out!
English
2
0
1
39
am.will
am.will@LLMJunky·
Codex Update 0.110.0 is here and it's F A S T 🏃 This release quietly introduced some big upgrades to Codex: > /fast mode > faster shell execution > multi-agent workflow upgrades > plugin + marketplace system > better memory Here’s what’s new 👇
am.will tweet media
English
29
23
425
31.1K
Alexander Kim
Alexander Kim@avxkim·
@LLMJunky maybe there's another feature flag for enabling marketplace? :)
English
1
0
1
41
am.will
am.will@LLMJunky·
5/ Plugin Marketplace Codex now has a plugin marketplace. Plugins can provide: • Skills (prebuilt workflows) • MCP tool servers • App connectors They can be loaded from config or a local marketplace. Even more interesting: the app server now includes an install endpoint for plugins. This will make installing skills, MCPs, and Apps incredibly simple.
English
7
0
20
2.4K
Alexander Kim
Alexander Kim@avxkim·
@melvynx way faster & and way dumber honestly, observing this with 5.3-codex xhigh
English
1
0
1
80
Melvyn • Builder
Melvyn • Builder@melvynx·
Codex is way faster and more enjoyable to use !
Melvyn • Builder tweet media
English
14
0
99
8.5K
Alexander Kim
Alexander Kim@avxkim·
@haider1 $100 plan for codex only with same codex limits $200 plan currently has. I don't use 5.2-pro, image generation, sora, etc
English
0
0
0
191
Haider.
Haider.@haider1·
if openai introduces a $100/m plan it should offer better value than both the $20 and the $200/m it should include: > access to stronger models and a larger context on chatgpt web > 3-4× higher codex usage, with a larger context if possible if not, the $100 plan doesn't make much sense
English
20
1
74
12.2K
Alexander Kim
Alexander Kim@avxkim·
@LLMJunky In terms of Code i wouldn't say that 5.3-codex xhigh is miles ahead of Opus 4.6, i feel like they're close to each other.. But so much noise about codex these 2 months lol
English
1
0
0
28
am.will
am.will@LLMJunky·
@avxkim ive gotten some pretty good uis out of codex but yeah opus.
English
2
0
1
132
am.will
am.will@LLMJunky·
🥶 Themes in Codex? There's a little something in here for everyone. All your favorite themes get a compatible color schema. *Or* Just make your own ~/.codex/themes Cooking 🧑‍🍳 Will be included in update 0.105.0*
English
5
2
105
6.9K
Alexander Kim
Alexander Kim@avxkim·
@LLMJunky i've tried many things, tried to feed it web-design philosophy of our famous local web-designer - didn't help, still was producing that sloppy UI... So i just re-subscribed to CC pro, hit the limits, re-subscribed to Max now and back to back comparing opus 4.6 to 5.3-codex xhigh
English
0
0
1
17
Chong-U
Chong-U@chongdashu·
Codex CLI will have MULTI-AGENT support in 0.123.0 Here's how it will work: → 1. Add `multi_agent=true` in ~/codex/config.toml under [features] → 2. Toggle with `/experimental` → 3. Three built-in agents (default, explorer, worker) → 4. Define your own agents ``` [agents.researcher] description = "Research-focused role." config_file = "~/.codex/agents/researcher.toml" model = "gpt-5.3-codex" model_reasoning_effort = "high" web_search = "live" sandbox_mode = "read-only" developer_instructions = """ You are the researcher agent. - Gather evidence from primary sources. - Return links, findings, and uncertainty. - Do not edit files. """ ``` → 5. Run more agent threads per session (default: 6) ``` [agents] max_threads = 6 Notes: → OAI released 0.122.0 - but rolled it back! → I built it from source (so you don't have)
English
14
21
307
36K
Victor
Victor@VAdatsi·
@PaulSolt i am a pro user on codex . $200 . just so you know . and you are very right , they have different strengths
English
1
0
1
17
Paul Solt
Paul Solt@PaulSolt·
Both are good. Codex is better. With limited time, I’m all-in on one. Anthropic models are bad coders unless you like to babysit. Claude is too much extra work. I have kids and limited time. I’ll babysit kids, but not agents. No thank you!
Marcos Saraiva@msaraiva

You can easily spot people are shilling for a company when they make claims like these. Both tools each have their strengths and anyone smart enough is using the right one at the right time.

English
14
2
68
7.6K
🍓🍓🍓
🍓🍓🍓@iruletheworldmo·
OPUS 4.6 IS LIVE.
🍓🍓🍓 tweet media
English
46
23
798
55.2K
Paul Solt
Paul Solt@PaulSolt·
@camsoft2000 Currently, GPT-5.2 is not good at UI Claude can fix these issues. Helped a ton with my spider app.
English
4
0
6
631
camsoft2000
camsoft2000@camsoft2000·
Using Codex but hate its UI/UX design skills, ask Claude Code to redesign the UI. Thank me later. My method: - Build MVP with Codex - Redesign with UI with Claude Code
English
14
1
39
4.6K
nic
nic@nicdunz·
i want gpt-5.3 now please
English
21
11
295
14.6K
Alexander Kim
Alexander Kim@avxkim·
@PaulSolt So you're gonna switch to this app? Honestly CLI is fine to me
English
1
0
1
443
Paul Solt
Paul Solt@PaulSolt·
The Codex app feels like the missing IDE layer for the Codex CLI. One-click to open Xcode, run iOS/macOS apps, review diffs, and manage agents. Hard to go back after this.
English
18
25
328
49.8K