π Gemma 4 on SwanChain.
β 31B is live now
π 12B arrives June 10
Optimized for:
π AI Roleplay
π€ Digital Humans
π€ AI Companions
π’ AI Employees
Run powerful Gemma 4 models with lower VRAM requirements and production-ready inference on SwanChain.
https://t.co/Nvou751Cgk
#Gemma4 #SwanChain #RoleplayAI #DigitalHumans #AIAgents
Gemma 4 can run on phones without an internet connection! π€―
It can perform local agentic tasks, such as logging and analyzing trends. When connected, it can also make API calls.
Want to try it yourself? Get the Google AI Edge App on iOS or Android. (π Sound on for the demo!)
π³οΈ SIP-003 vote ends tomorrow.
β FOR = transition to Swan 2.0
β End UBI, providers earn 95% in stablecoins, hardware tiers, 20% SWAN discount
β AGAINST = keep current UBI model
β Emissions continue, no hardware requirements, no inference revenue
Vote: https://t.co/KAAmkwijAJ
What's live at launch:
β OpenAI-compatible API at https://t.co/bLkIbAxSa4
β 8B to 70B model catalog
β Pay with SWAN (Swan Chain mainnet) β 20% discount
β Pay with USDC (Ethereum mainnet)
β 95% of all fees go to GPU providers in stablecoins
β Free tier: 50K tokens/day, no KYC
π’ Swan 2.0 is Live β computing-provider v0.1.0 Released
Today marks the last day of the legacy flat UBI. Starting tomorrow, Swan 2.0 contribution-weighted rewards go into effect.
What this means:
Providers serving real inference earn USDC/USDT directly (95% revenue share)
Idle GPUs earn zero β no more flat rewards for 0% uptime
The UBI taper begins: 100% β 50% (April 10) β 20% (May 10) β 0% (June 10)
The new software is out:π computing-provider v0.1.0 https://t.co/SDjyqvqFkY
This is purpose-built for AI inference serving on Swan 2.0 β if you have a qualifying GPU, now is the time to get set up while UBI is still running as a safety net.
The transition timeline:
β March 16: Swan 2.0 launches, inference API goes live β April 1β7: Community governance vote on UBI sunset β April 10: UBI begins tapering (50%) β May 10: Down to 20% β June 10: UBI off permanently
Full SIP-003 proposal is live. Read it. Vote on it. This is your network.
https://t.co/KL8zyaHdDK
Swan 2.0 is live on March 16.
A full redesign of how Swan works:
Market-driven inference marketplace replacing UBI subsidies
Tiered model catalog with per-token pricing
95% revenue directly to providers β in stablecoins
Pay-with-SWAN for 20% off inference
OpenAI-compatible API from day one
For SWAN holders:
UBI was emitting 21.3M SWAN/year in sell pressure. Providers had to dump rewards to pay their bills.
SIP-003 stops that entirely.
Token utility becomes real: Pay-with-SWAN for 20% off inference. Collateral. Governance. Demand-driven, not emission-driven.
Introducing SIP-003: Swan 2.0 Inference Cloud
The core principle is simple:
GPUs that don't serve real inference workloads don't earn SWAN rewards.
No more flat UBI. No more rewarding idle registration. Only contribution earns.
The numbers don't lie:
58,369 SWAN/day in UBI emissions
$21.95/day goes to providers doing zero work
Best active provider earns $2.63/day β not enough to cover electricity
0 operational 70B GPUs despite 73 being "registered"
We need to be honest with you.
Swan 1.0 has 166 registered GPUs across 64 providers.
Only 6 are actually operational. 75% of daily SWAN rewards go to GPUs with 0% uptime.
That's not a network. That's a subsidy program for idle hardware.
Time to fix it. π§΅