stoikol

@llmdevguy i agree, especially i highlight that it is a great package for "inspecting" and tinkering, but actually for production it is not very operational, since the monthly and weekly limits are quite restricted. additionally, the agent's inference speed is very very slow

🚀A short review after a full month with OpenCode GO. TDLR: this is the best-value coding subscription I've used. The model selection is excellent, and at this price point, I don't think there is a better option on the market. 👉There is only one serious drawback: the monthly limit. If there were no monthly cap, the service would be close to perfect. Unfortunately, the monthly allowance is only slightly higher, than the weekly one. Take a look at the screenshot. A 5-hour limit allows you to get a lot done. During that time, I used 20% of my weekly quota. That's not bad, but the issue is that it also consumed 10% of my monthly quota. That is why this plan is ideal for: ⭐️Light usage ⭐️Complementing ChatGPT Plus ⭐️Serving as the brain for Hermes agent Overall, it's a fantastic product. If you haven't tried it yet, the first month costs only $5, and then $10 per month after that. An excellent deal.

@moonpay moonpay pushing crypto via ai adoption !! follow congrats guys

BREAKING: MoonPay has acquired Dawn Labs

@CopyRebeldia soo cool

Alguien acaba de fulminar a la industria de visualización urbana 3D. Repo en GitHub. Le metes una ciudad y te escupe el modelo 3D con edificios y calles desde OpenStreetMap. Exporta GLB. MIT. Sin login, sin API key.

@declan_lw thanks for sharing, i will explore your arbitrage arbitrage, im really interested to see how are you matching events cross paltforms

@KobeissiLetter holy moly

This is absolutely insane. President Trump is currently flying to China with all of the following people to request "deals" with China's President Xi: 1. Elon Musk, Tesla and SpaceX CEO 2. Jensen Huang, Nvidia CEO 3. Tim Cook, Apple CEO 4. Larry Fink, BlackRock CEO 5. Stephen Schwarzman, Blackstone CEO 6. Kelly Ortberg, Boeing CEO 7. Brian Sikes, Cargill CEO 8. Jane Fraser, Citigroup CEO 9. Larry Culp, General Electric CEO 10. David Solomon, Goldman Sachs CEO 11. Sanjay Mehrotra, Micron CEO 12. Cristiano Amon, Qualcomm CEO President Trump also says there are "many other" CEOs joining him on the trip who have not yet been disclosed. Never in history has such a trip even remotely near this scale and caliber occurred. This Trump-Xi meeting is far bigger than most realize.

@intheworldofai i would like to see different skin themes like old winamp era. if i have some time i will try to develop one

Hermes Agent is evolving FAST. In just the past week, Nous Research added: - A full WebUI/Desktop App - Background Computer Use on macOS - Multi-agent orchestration - Hermes Kanban upgrades - Lightpanda browser backend support - Qwen3.6-Plus FREE in Nous Portal - Better autonomous workflows - Persistent long-term memory systems Hermes is starting to feel less like an AI tool and more like a true open-source Agentic AI Operating System. Full breakdown/demo: youtu.be/Gx2joHxUhgg

@datadashboards okay noted for research! thx

@air_codex depends on what you or your users will want if its a product for others. limitless is good for short term crypto markets / football. hyperliquid only has one recurring and one type of market right now. predict is more of a mix of everything

Prediction Market Notional Volume Recap (2026-05-04 vs 2026-04-27) 2026-05-04: 7.2B | 2026-04-27: 7.0B • Kalshi: 4.1B (+8.5% WoW) • Polymarket: 1.6B (-6.2% WoW) • Limitless: 600.4M (+20.8% WoW) • Polymarket (US): 298.9M (+9.6% WoW) • Predict: 232.4M (-23.5% WoW) • Opinion: 207.0M (-7.6% WoW) • Crypto(.)com: 161.3M (-2.8% WoW) • IBKR: 5.9M (-5.3% WoW) • Overtime: 4.9M (+14.6% WoW) • Myriad: 625.5K (-29.4% WoW) • Total: 7.2B (+3.7% WoW)

@Gsnchez thermodynamics sir!! the mother framework

Nadie me creería si os dijera que lo primero que he hecho es informarme sobre topología y mecánica de fluidos (ésta última la aprobé con un 6 cuando hice la FP de Química Industrial con 21a). Llevo tiempo pensando en arquitecturas multi-agente aplicadas a gestión patrimonial y me convenzo de que el marco mental correcto no viene de la ingeniería de software clásica sino de la matemática de sistemas continuos. Una cartera no es un orquestador llamando a herramientas: es un campo de creencias distribuido sobre un manifold de activos donde cada analista, cada modelo y cada señal ocupa una región con curvatura propia. Topología te da la noción de qué regímenes son contiguos y cuáles requieren un salto. Mecánica de fluidos describe cómo se propaga la información entre clases de activo cuando cambia el régimen macro. El free energy principle de Friston explica por qué un comité bien diseñado minimiza sorpresa actualizando creencias en lugar de promediar votos. Optimal transport da una distancia real entre dos asignaciones de cartera (Wasserstein, no euclídea). Kuramoto explica por qué a veces los analistas se sincronizan justo cuando no deberían. El stack actual de agentes LLM con tools y memory es el ensamblador de esta capa: funciona, pero el lenguaje de alto nivel seguramente todavía está por escribirse, y va a ser geométrico antes que sintáctico.

@datadashboards I'm fetching data from Poly and Kalshi to create tools for oracle study and money flow. My next plan was to integrate the Hyperliquid API. Now, looking at your chart, I wonder if I should integrate Limitless or another provider

@air_codex Hmm thanks for flagging mate. will take a look later

Built a lot and ready for more! I'm looking for a job, please help spread the word. Thanks! 🙌

@Gsnchez good luck my sir

Me voy a meter en el proyecto más difícil al que me he enfrentado hasta la fecha, otra producto cutting edge más único en España. Deseadme suerte. Si lo logro cambio la industria.

@yevhenx thx sir!! you rock

@air_codex for time-series/historical charts - recharts and for alluvial - custom-built using native React <svg> elements

preview of the ethereum lending dashboard -aave, spark, morpho, fluid, euler coverage -vaults, rates, allocations, flows, risks -history for all the individual positions -unlimited api access for real-time monitoring everything is ready, public release next week

@PredictParity sent 🤝

@air_codex send us a dm

See top traders for any polymarket. Live on all market pages.

@vmiss33 Great thread! You should tell your Lifestyle Agent this: "Every 30 minutes while I'm working or sitting, gently remind me to stand up and do full-body dynamic movements covering all muscle groups for at least 3 minutes."

@mddanishyusuf wow sir, was not 2,5k so cheap deal?? you are killing it

For sale: crawlapi.dev Price: $2,500 Age: 1 month Users: 31 MRR: $0 Good domain Solid product. I just don't have bandwidth to market it. Looking for an owner who does. First come first serve. Here is link to buy: buy.stripe.com/5kQ9AT3qOfGDeG…

For the past 8 months I've been working on a side project called Toroot, a prediction market built around the idea of skin-in-the-game & collective knowledge. Today I've decided to pause it. I want to share what happened, what I managed to build, and where I'm headed next. 🧵

@vmiss33 @NousResearch yeah that's exactly what i meant dear madame, with deepseek, it would be that "inexpensive" probably x5 times cheaper than that $10 monthly cost x.com/neural_avb/sta…

@air_codex @NousResearch I’m all about doing it as inexpensive as possible, but yes you could do that too

Right now my Hermes setup is 4 agents with providers as follows: 1) @NousResearch portal (10 dollar a month plan) 2) local model (free, Qwen4.5-9B quant on an RTX4070) 3) open router (free models only but I put in 10 bucks to get more requests) 4) nvidia nim (free only) bonus: burst up to @HotAisle where I can rent a MI300x for 1.99 an hour billed by the minute for intense stuff (and I had hermes build a skill for me to handle it). Are there trade offs? Yes, but don't not do this because of $$$. The agents themselves run on a super old intel nuc I had laying around. you can do this on the cheap.