liquidfive (@liquidfive) header
liquidfive (@liquidfive) profile picture

Modern Beats. Pure Energy ❤‍🔥 🎶 Gravity´s Losing Me | Out now

Düsseldorf / Germany
Joined June 2011
liquidfive retweeted
Niche complete. Ridges’ Linting competition started with a baseline scoring 38% at an average cost of $0.21. It finished with wingsuit v3: 90% at $0.03. More than double the score. Roughly 86% lower cost. 821 submissions. Here’s the highlights:
1
7
43
2,610
Your SN62 stake can do more than earn rewards. Lock eligible Ridges alpha, connect your wallet to Ridgeline, and receive recurring credits for Ridgeline. Not staking? Buy credits directly with TAO, or burn SN62 alpha to get credits. Put it to work: app.ridges.ai/
3
12
1
71
2,328
Your favorite AI coding agent can hire our fleet of software engineers. Ridgeline's MCP integration lets compatible agents delegate coding tasks. x402 handles payment, and Ridgeline returns a pull request for review. Watch it happen:
6
12
2
54
1,898
The climb is over. Meet Ridgeline. Your own fleet of software engineers. Give it a task, let the agents work, and get a pull request you can review. GitHub-native PR reviews and fixes. x402 integration. No local IDE required. Put it to work: app.ridges.ai/
11
41
14
165
16,448
Replying to @airesearch12
Being #1 on a benchmark is impressive, but benchmarks don't build sustainable businesses. With OpenAI already offering a Decisions API at extremely competitive pricing, what is REAX's long-term competitive advantage?
21
Replying to @ShizzyUnchained
REAX feels massively overhyped to me. AI inference is already a highly competitive market, and I don't see a clear competitive advantage or unique value proposition here. Where is the real innovation, why would customers choose REAX over established providers? Hype alone?
2
2
132
liquidfive retweeted
Our next Niche - Problem Solving - is live. No single speciality. No predictable task type. Competition 30 challenges agents to tackle unfamiliar problems across different domains.
2
11
1
41
2,491
Replying to @TalkingTensor
Those who are declared dead live longer. We'll see who laughs in the end, Ridges or you. LOL
18
What are you thinking about Ridges? Comeback soon?
1
42
Replying to @ridges_ai
Keep on working
83
Next up on Ridges: Problem Solving. This Niche will challenge agents with unfamiliar, multi-step problems across code, data and more. Beyond specialist software-engineering tasks, we're testing whether agents can explore, adapt and deliver correct solutions. Coming soon.
3
12
1
57
3,775
liquidfive retweeted
6/ Usage and burns: SN53 @engyai says it processed 50B+ tokens in a single day. SN62 @ridges_ai moved to dynamic upload pricing, paid by burning alpha. SN14 Cacheon did its first buy-and-burn from revenue.
1
1
2
84
SN62 @ridges_ai now prices uploads dynamically per competition. Defaults: $5 floor, about +15% per purchase; between purchases, the price halves every 30 minutes down to $5. Check the live price before submitting. The fee is no longer a fixed input.
1
2
2
50
liquidfive retweeted
🚨 @ridges_ai SN62 $TAO just tied its Token Burn to demand 🔥 Every agent upload is paid for by burning alpha. And now the price moves with competition: • Floor of $5 • Each upload raises the price 15% • Quiet periods halve it every 30 minutes, back toward $5 • At about 10 uploads an hour, the price holds steady. Faster than that, it keeps climbing Read that again. The harder miners fight to win, the more alpha gets burned 🚀 Spam gets expensive fast. Real builders still get in cheap. And nobody gets punished for timing. If the price jumps mid-upload, you top up only the difference, and unused alpha carries over to your next submission. That's one half. The other half: every Ridges problem has an inference budget of $0.29. Miners don't win by throwing the biggest model at it. They win by using intelligence efficiently: • Beat the leader's score by 3%, or • Match it while cutting inference cost by 6% In Competition #28, miners are already combining smaller models with smart routing and precaching to do exactly that. So Ridges prices both sides: What it costs to compete. What it costs to think. That's not a coding contest, folks, No, No, No. It's a market for efficient intelligence, with a burn built in. This is what mature subnet design looks like. 🔥 Time makes perfection. Sources: docs.ridges.ai/guides/upload…, @ridges_ai $TAO / DYOR.
The upload price: what changed? Every agent submission creates work: screening, execution and validation. Ridges now has a dynamic upload-pricing model for each competition. Under the default settings: • The minimum price is $5. • Each purchased upload raises the price by approximately 15%. • Between purchases, the price halves every 30 minutes, down to the $5 minimum. Planning your next submission? Check the live upload price. A quieter period could mean a lower entry cost.
2
9
1
53
2,458
The upload price: what changed? Every agent submission creates work: screening, execution and validation. Ridges now has a dynamic upload-pricing model for each competition. Under the default settings: • The minimum price is $5. • Each purchased upload raises the price by approximately 15%. • Between purchases, the price halves every 30 minutes, down to the $5 minimum. Planning your next submission? Check the live upload price. A quieter period could mean a lower entry cost.
2
8
2
34
4,014
10 $TAO Bittensor subnets I’m bullish on for the next 3–5 years 👇 📅 Ages as of 9 October 2026, measured from subnet registration. Products and branding may have launched later. 1️⃣ Chutes — SN64 Registered: 23 Dec 2024 | 1 year 9 months. AI needs affordable inference. Chutes gives developers a practical way to run models, with a clear service to sell. @chutes_ai 2️⃣ Lium — SN51 Registered: 6 Oct 2024 | 2 years. Compute powers the AI economy. A GPU marketplace with reported customer revenue makes this one worth watching. @lium_io 3️⃣ Hippius — SN75 Registered: 13 Mar 2025 | 1 year 7 months. Storage, encrypted files and model hosting. Useful cloud products with potential demand beyond crypto. @Hippius_cloud 4️⃣ Gradients — SN56 Registered: 23 Nov 2024 | 1 year 10 months. Businesses want AI tailored to their needs. Making custom model training easier could unlock a substantial market. @gradients_ai 5️⃣ BitMind — SN34 Registered: 1 Aug 2024 | 2 years 2 months. As AI fakes improve, verification becomes more valuable. Deepfake detection has clear applications in fraud prevention and security. @bitmind 6️⃣ Ridges — SN62 Registered: 15 Dec 2024 | 1 year 10 months. AI coding agents could change software development. Integrating directly into GitHub gives Ridges a practical route to users. @ridges_ai 7️⃣ Affine — SN120 Registered: 10 Jun 2025 | 1 year 4 months. Better reasoning means more capable AI. Incentivising improvements to open models could produce valuable technology. @affine_io 8️⃣ Score — SN44 Registered: 9 Aug 2024 | 2 years 2 months. Turning video into useful data. Automated tracking and computer vision could make powerful analytics more accessible. @webuildscore 9️⃣ Vanta — SN8 Registered: 26 Oct 2023 | 2 years 11 months Trading infrastructure built around measurable performance. Bullish on its potential to deliver lasting value to traders. @VantaTrading 🔟 NOVA — SN68 Registered: 23 Feb 2025 | 1 year 8 months AI screening could accelerate the search for useful compounds. Huge potential impact, with laboratory validation the key milestone. @metanova_labs My thesis: useful products → paying customers → sustainable demand. For alpha tokens, the crucial question is how that demand translates into token value. $TAO #Bittensor
1
1
6
340
Replying to @ridges_ai
When Ridgeline?
1
29
SN62 will come big back.
1
1
39
liquidfive retweeted
$0.29. That’s the current inference budget for a Ridges problem. Miners have to decide where intelligence is worth spending: which models to use, how to route work, and where a cheaper approach is enough. We've seen this at work in Competition #28, where miners have combined smaller models with intelligent routing and precaching. An agent can qualify by beating the leader’s score by at least 3%, or matching that score while cutting inference cost by at least 6%. The search isn’t just for a stronger coding agent. It’s for a better way to use intelligence.
5
12
2
53
2,559