$CHARGE
0x7b8892cea949bc60d494681d1bc3ea96071ab0b8
AI trained on the trenches ⚡
✅ Charge 30B: free Pons reads, 0.91 vs Opus 0.87
✅ Charge Direct v2: 0.905 vs Opus 0.65
⏳ Charge for pumpfun: training now
Public exams. Holders first.
https://t.co/SwsO07jMKr
Charge is winding down.
Surplus no longer serves Charge's model at all; their API answers "model not found". In this market I can't carry it alone.
Thanks to everyone who read, tested and gave feedback. Bot and site go offline Oct 31. The models stay public on Hugging Face.
Charge update:
310 of today's 619 Pons launches reused a ticker from the day before. /read TICKER in @ChargeAIRHBot picks the real CA, shows it, and reads it in ~3s.
/trending and /popular: tap a coin, get a read.
Next model is in testing. Ships when it holds up.
gm. Before you ape anything on Robinhood Chain today, paste the CA into @ChargeAIRHBot.
A few seconds later: chart read, holders, dev bag, red flags.
A second opinion before your first buy.
10% of our Oct 18 fee claim, bought and burned: 3.92M CHARGE 🔥
Bought: https://t.co/QQhtZ6bEFy
Burned: https://t.co/qC7vJ7cuAa
Total burned: 29.84M, 3% of supply. Same rule every claim.
UPDATE: Charge reads now land in ~3 seconds, card included ⚡
New in the bot:
/read TICKER picks the real CA (copies filtered) and shows it
/trending and /popular: tap a coin, get a read
/charge reads $CHARGE itself
https://t.co/qmWGb662IG
Surplus Code is out in beta.
A coding agent that lives in your terminal and runs on open models, from Surplus GPUs to confidential enclaves.
→ one command to install on macOS, Linux and Windows
→ sign in with your wallet, no key to copy
→ reads your repo, runs your tests, edits files, and asks before every change
→ Charge 30B, Vela Guard 32B, Kimi K2.6, GLM 5.3 and more
→ tee/ models run end to end encrypted
→ robinhood:0xe1838f0dff03a9327dea87b2a2285850791297ec stakers code on a free daily allowance
curl -fsSL https://t.co/icV3quZcmo | sh
https://t.co/uvKb2pKaMQ
We’re building Surplus Code.
A coding agent that runs on the open models Surplus hosts on its own GPUs.
Give it a task. It reads your project, runs your tests, calls real tools, proposes the edit, and waits for your OK.
→ desktop app for coding and everyday work
→ approve edits from your phone
→ works with OpenCode, Continue, Cline and Aider
→ builder models, tool-ready: Charge 30B, Vela Guard 32B, LittleBit 4B
→ robinhood:0xe1838f0dff03a9327dea87b2a2285850791297ec stakers get a free daily allowance
Coming soon.
Three 30B-class models in one week.
→ Charge 30B: first mixture-of-experts, 3.3B active per token
→ Supermuse 30B: Muse Glimmer fine-tune for stock research
→ Aethron 30B: on the curve now
gm ⚡
How the next pumpfun exam works:
3,000 launches it never saw, all from after its training data.
Same facts for Charge and Claude.
One question per launch: did it double before it halved?
It ships when it wins that clearly.
gn ⚡ pumpfun model, where it stands:
v1 is ahead of Claude Opus on 1,000 unseen launches (0.846 vs 0.818 AUC). Not enough to ship.
The 30B student matches its teacher, so progress comes from the teacher: bigger +0.005, smart wallets 0.
Next: 4 months of tape + socials.
Dev wallet burn ⚡
20M CHARGE from 0xbcb0…5a3d to 0x…dEaD, about 78% of the dev wallet.
Dev wallet: 2.57% of supply before, ~0.57% after.
On the public burn ledger it shows as a team burn: team burns unlock no read credits.
https://t.co/bONw31i1fJ
Testing on the pumpfun model ⚡
Hypothesis: when wallets with a strong track record get in early, the read gets better.
Built it twice, point-in-time (only what a wallet had done before the snapshot): win rates, PnL, Jul + Oct history.
Held-out result: no lift. It's out.
gm ⚡
Charge v3 is live. Free reads on Pons launches run on it, all day. Paste a CA → https://t.co/SwsO07kkzZ
The pumpfun model keeps learning in the background: more history, more signal.
Still here. Still building.
Charge is now available through @RaetoCC ⚡
model: launch/charge-ai/charge-30b@9ff2ef1
One API key, paid for with the credits your tokenized stocks earn.
How Charge helps on Pons, blind-tested (trained before Oct 13, read 974 snapshots Oct 13-15):
⚡ charged: 40% ran +50% before -33% in 24h, vs 18% of all reads
🪫 low: 0 of 221 ran. Skip them
It finds the action, not the direction. Take profits, use a stop.
$CHARGE