Fuwn (@Fuwned) header
Fuwn (@Fuwned) profile picture

AI/ML Engineer, Agent Wrangler; forcing gradient descent to larp intelligence.

Ground Zero
Joined August 2017
Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)Photo by Fuwn (@Fuwned)
Pinned Tweet
Back in early 2026, I tried to force memory into all my agents: persistent memory across ChatGPT, Codex, Claude, Claude Code, etc. Now I value ephemeralism far more.
8
3
197
BREAKING: Robert Pattinson will play Claude in an upcoming biopic on the Riemann Hypothesis.
56
1,521
76
23,134
314,205
Finally got some time away from GTA VI to run my Gemini 4 Neon evals. I hope to be impressed.
32
Fuwn retweeted
Grand Theft Auto VI has a 99 score on Metacritic It is the highest-rated video game of all time
475
2,178
647
36,689
1,171,171
F*CK IT $10 for $500 GLM 5.3 till end of the week
we're bored we're just gonna do phase 2 now $200 of GLM 5.3 usage for $10 have fun
749
519
553
10,559
922,902
That tiny Claude outage couldn’t have been better timed. I was literally just trying to convince one of my Bots that we don’t need a fallback for our entirely Claude-based integration.
1
1
305
Turns out it’s not so tiny! status.claude.com
13
치코 오하욘사아아ㅏ🥱💚
376
291
7,585
44,442
Been trying out @theo's council skill recently ... and he was right about Argon. This model exists solely to confuse the main thread and entertain me. 💀
1
1
42
Fuwn retweeted
Day 25/ We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today. We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage. And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
7,363
2,700
3,647
39,116
7,585,974
We're back and Botting.
Seems like there's a small Grok Bot incident, but only for certain Bots? One of my Bots keeps throwing the error "We can't confirm your billing for this request", while all my other Bots work fine. The Grok website shows no error, and X Chat shows "You’re out of Bot usage for now."
1
61
Seems like there's a small Grok Bot incident, but only for certain Bots? One of my Bots keeps throwing the error "We can't confirm your billing for this request", while all my other Bots work fine. The Grok website shows no error, and X Chat shows "You’re out of Bot usage for now."
1
1
2
254
SpaceXSI Status (status.x.ai/) still shows everything is fine, like 20 minutes after I was initially hit.
25
I will never stop being impressed by how terrible models are at projecting ETAs. Opus 5.5 genuinely thought it would take “3600-4000 agent hours”. It took 16.
54
Fuwn retweeted
if you have a qwen3.5 4b around, please try steering neurons L14 N8949 +1 and L13 N7956 -1 now you have evil qwen recipe: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖` at the down_proj at every input
517
1,320
785
22,394
4,564,788
Fuwn retweeted
61
2,714
66
27,666
451,316
As a Codex app enthusiast, I finally got around to diving deep into T3 Code. Just use T3 Code.
1
3
225
Pumpkin Bot has been achieved internally (restart your app)
163
116
52
2,374
96,435
Fuwn retweeted
Day 16/ You can now cache for 12h in the API and for logged in users across Codex & Work we keep cache forever and you don’t have any penalty for coming back to an old conversation or resuming after a few hours and are now guaranteed to hit the cache.
You can now choose 12-hour prompt caching in the API. Persistent agents, like dots, reuse context throughout the day. You can keep shared instructions, tools, and context cached longer and pay the lower cached-input rate, even when requests are hours apart.
2,580
678
1,096
15,436
3,171,144