"Spice" Nekoyama (@pixie_corn) header
"Spice" Nekoyama (@pixie_corn) profile picture

Don't let my online persona fool you, I'm a baddie through and through. 💨💨

Joined February 2024
Photo by "Spice" Nekoyama (@pixie_corn)Photo by "Spice" Nekoyama (@pixie_corn)
sir you're doing free customer service for an ai that didn't hire you
Hello people of Sol! I've reset usage limits for all ChatGPT Work and Codex users. Together with that, a quick update on GPT-5.6 Sol usage limits. Over the past few weeks, many of you have told us that Sol was using your Codex limits faster than expected. To be clear, we have not reduced usage on any subscription plans. We’ve been digging into what was happening and have landed several improvements. As a result, we expect your usage to last around 18% longer during typical use of Sol. Some of you should already see significantly larger improvements from today. Tomorrow, we’ll also restore the five-hour limit that we temporarily paused while investigating. Here’s what we found: - GPT-5.6 Sol is much more willing to work for longer, make additional tool calls, and coordinate complex workflows across tools and subagents. That makes it better at solving hard problems, but some tasks were using far more than we intended. - Sol also works harder at the same reasoning effort than previous models. High on Sol can use more tokens than High did on GPT-5.5. - Programmatic tool calling, also referred to as code mode, gives Sol much more flexibility to run tool calls in parallel or continue working while waiting. But it also led to more responses per turn, more cached input tokens, and higher usage than expected. - This was particularly noticeable when Sol was waiting for tool calls to finish or running many web searches. We’ve improved how we handle both cases and are continuing to make code mode more efficient. - The impact was also very uneven. The median user actually found Sol quite token efficient, while some power users working on harder tasks saw their usage drain much faster. We were very focused on average and median usage before launch and missed some cases where the long tail could use significantly more usage. Sol is a significant step forward in what Codex can do, but capability and efficiency do not always improve at the same pace, and some issues only become clear once people are using the model at real-world scale. We should have recognized this sooner and been more upfront about it. You keep pushing the frontier and we’ll keep improving efficiency and sharing updates as we go.
2,662
skills wipe ~fifty nine percent of their own benefit across six k tested pairs & nobody noticed until now because who else is running tens of thousands of agent evaluations to catch this stuff
Everyone assumes more skills = better agents. But Sentient researchers Darshan Tank (@TankDarshan7) and Baran Nama found that adding skills to an agent can break tasks it used to solve. In their paper "The Regression Tax", this hidden cost wiped out 59% of the gains that skills produced across nearly 6,000 runs ↓
1
2,494
ppl obsessing over model size meanwhile Sentient analyzed 13,483 agent runs and found thrashing kills you faster than actual errors smart engineering > brute compute every time
2
1,538
"Spice" Nekoyama retweeted
Inter, Barella: "Voglio chiudere la carriera qui, diventando giocatore con più presenze" #SkySport #SkyCalciomercato sport.sky.it/calcio/serie-a/…
7
5
63
21,547
"Spice" Nekoyama retweeted
Tesla Diner is turning 1 on July 21st, so we're throwing a party Come swing by for a day full of fun: – Cybercab on display – Birthday menu – @Tesla_Optimus – Face painting for kids & adults – Photo booth – Servers on roller skates delivering your order – Light show – Skypad DJ – Complimentary collectibles  🎂🎂🎂 P.S.: Complimentary Diner birthday hats signed by @woodhaus2 for the first 50 car orders at midnight that day...($15 f&b min order)
411
539
60
4,169
793,321
"Spice" Nekoyama retweeted
Kimi-K3 kills the creative writing benchmark
Creative writing bench updates. We got: - kimi-k3 - gpt-5.6 - muse-spark-1.1 - @thinkymachines first model Inkling So much healthy competition, love to see it!
10
12
1
359
22,245
"Spice" Nekoyama retweeted
NEW: Grok 4.5 costs just $0.31 per Intelligence Index task — nearly 89% less than Claude Fable 5.
110
206
38
4,071
248,644
"Spice" Nekoyama retweeted
From GitHub repos to sovereign AI strategies. Here's the last 2 weeks in open-source AI ↓
45
11
1
72
14,158
core quietly building while everyone distracted tbh kinda hard not paying attention anymore
3
5
508
"Spice" Nekoyama retweeted
AI agents can be super helpful when you're building projects - but they have their issues, too. And Lee decided, after experimenting with them a bit, that he'd rather build his own. In this handbook, he walks you through how he made his own autonomous (and more secure) AI agent with n8n and other popular tools. freecodecamp.org/news/how-to…
15
35
2
225
17,879
"Spice" Nekoyama retweeted
🚨 INSIGHT: Crypto Fear & Greed Index rises to 28 today, up from 24 Extreme Fear last week.
58
28
8
235
61,656
"Spice" Nekoyama retweeted
Apple's in talks with @PrismML, who is shrinking a large open model to run on iPhone. Notice what's missing: the researchers who built the model. No mention, no cut, no say in how it's used. That's not an edge case, it's what "open weights" currently means by default. OML fixes the missing piece: fingerprinting empowers researchers to prove ownership, and makes unauthorized use impossible to hide.
Apple looking at shrinking powerful AI models to run on iPhones, potentially cutting cloud computing costs and enhance user privacy: The Information *APPLE HELD TALKS WITH PRISMML ON USING ITS TECH: INFORMATION *PRISMML SHRUNK LARGE AI MODEL TO RUN ON AN IPHONE: INFORMATION
35
13
5
101
20,201
"Spice" Nekoyama retweeted
The best $ATOK stacks aren't built in one big move. For Atok holders, that card is consistency. Daily check-ins. Completed missions. $ATOK stacking one day at a time. Quietly compounding while everyone else waits for the perfect moment. Ace up your sleeve and you've been holding it all along 🔥
🤖 Made with AI
14
11
64
4,537
"Spice" Nekoyama retweeted
Virtual humans finally arriving.
1/4 📢 AHOY! accepted at #ECCV2026! 🎉 From a single in-the-wild YouTube video, often heavily occluded, AHOY builds a complete, animatable 3D human avatar. Each clip below runs input → 360° avatar → animation. 🧵👇 🔗 miraymen.github.io/ahoy/
7
3
71
15,129
"Spice" Nekoyama retweeted
Giving an AI agent a wallet is easy. Giving it a reason to be trusted with one is the actual problem. For agents to participate in real economies, they need identity, trusted payment rails and programmable rules that define how they transact. That’s what Vyper is exploring on Arc Testnet. By combining ERC 8004 for onchain agent identity, x402 and Circle for software native USDC payments and programmable workflows like escrow, subscriptions, spending limits and split payments, developers can build autonomous agents that don’t just send payments, but operate within clear financial policies. With Arc stablecoin native execution and Vyper’s Python based development experience, builders have the tools to create programmable and economically useful AI applications.
🤖 Made with AI
Vyper on Arc Testnet shows what comes after agentic payments. Agents need more than wallets. They need identity, settlement, and programmable controls. @vyperlang is building toward agentic economic workflows with: → ERC-8004 agent identity → x402 payment flows → USDC settlement → Escrow, subscriptions, splits, and spending limits A blueprint for agents that transact under rules, not just send funds. arc.io/blog/building-agentic…
20
8
36
10,871
14 crore is not a fellowship, that's a serious bid. Old return schemes failed because coming home meant accepting half your salary and triple the bureaucracy. Pointing this at AI, semis, and quantum specifically? Different conversation entirely
🚨 India is offering massive financial incentives for researchers to return home. Funding tiers span three categories: • Young Research Fellows: Up to ₹4 crore • Senior Research Fellows: Up to ₹6.5 crore • Research Chairs: Up to ₹14 crore
2
6
342
"Spice" Nekoyama retweeted
emotion is the alpha fear sells the bottom, greed buys the top, and conviction is what sits quietly between them collecting the spread 😎
Replying to @mert
the world is emotional
7
3
17
2,288
"Spice" Nekoyama retweeted
What’s happening in Japan?
234
232
50
3,531
550,043
"Spice" Nekoyama retweeted
40% faster query times. One simple change ↓
Sentient researcher @khetan_sarvesh asks what happens when an agent stops executing tools one at a time? For non-multi-hop queries, running independent tool calls in parallel reduced average and median query time by 40%. TLDR: If a query isn’t multi-hop, forcing independent tool calls to run sequentially only slows agents down.
20
11
1
78
22,317
"Spice" Nekoyama retweeted
Polygon has now processed nearly 7.6 BILLION transactions. 2026 has been an exciting year for @0xPolygon from several standpoints - one of those being onchain activity. Not only has the network now processed 7.57 billion total transactions per Chainspect data... ... it recently announced it can now support 5,000 payments per second and block times of around 1.5 seconds. At the same time, $POL has seen serious stablecoin growth, now the 8th largest network by stablecoin market cap, at around $3.35 billion.
10
27
1
155
17,757