Charge is an open 30B model that reads Pons launches on Robinhood Chain. It learned from launches and what happened to each one next. On launches it had never seen, it beat Claude Opus 5.5 at the question trenchers ask all day: will this double before it halves? In a fraction of the time (3.6× faster) and at 40× lower cost per read.
Charge builds the facts card from the public tape and reads it, live. ETH-paired Pons launches up to 3 days old.
A few reads a minute per person. Not advice.
Pons reads stay free for everyone. Link a wallet that holds CHARGE and Charge Direct, higher limits and an API key unlock. Need more than your tier gives? Burn CHARGE for read credits.
Sign a message with your wallet. It's free: not a transaction, nothing moves, no approvals. Charge checks your CHARGE balance live on Robinhood Chain.
You send an exact small amount of CHARGE to the burn address from the wallet you want to link. The burn proves the wallet is yours, and it becomes read credits.
Burn address 0x…dEaD . Send from a linked wallet; credits land within a minute. Burns are permanent.
Charge only verifies: it reads signatures, balances and burns on Robinhood Chain. Linking is a signed message, not a transaction; the site never asks your wallet to approve a transaction or a token allowance, and Charge never holds funds. Burns are sent by you, from your own wallet. Tiers count CHARGE held in linked wallets (LP positions don't count) and follow live balances. Reads are not advice.
These are Charge's real reads of launches it never saw in training (October 10–12). We picked them to show every kind of outcome, misses included. The full numbers are further down.
Same facts, same question: will this launch double before it halves in the next 24 hours? Scored on launches none of them had seen.
How we tested. snapshots of live Pons launches from October 10–12, none seen in training; every launch that doubled first is included. Each model saw the same facts. The Claude models also got a primer on how Pons works and the 6% base rate, and rated 20 snapshots at a time. AUC weights each launch once. Test numbers come from Charge 30B trained only on launches before October 10; the released model is the same recipe trained on every launch through October 12. Check it yourself: every card, every model's answer, the outcomes and a scoring script are public on Hugging Face.
Charge ranks a launch against every live launch it learned from: 90% means busier and stronger than 90% of them. Here is what launches in each band did over the next 24 hours, from Charge's own reads on the test days.
of ⚡ reads halved before they doubled; doubled first. The typical one dropped to of the read price within a day. Charged is a reason to watch, not to chase.
Test days: October 10–12, 2026, one launchpad. Small samples, and markets change.
Public trades of every ETH-paired Pons launch since October 5 that got at least one real trade (about a quarter never do): price, flows, buyers, sellers, holders, the dev's own trades, ticker copies, socials. Nothing private.
Each launch read at fixed ages and every 10 minutes while it traded: 65,542 moments, each labelled with what happened in the next 24 hours.
A gradient-boosted model learns the odds from those labels, with every launch weighted once so a long-running coin can't dominate.
Every answer is written from the data by templates: the band, the measured outcome mix and the facts that moved the odds for launches like it. No text from another LLM.
Qwen3-30B-A3B fine-tuned with LoRA (attention and experts) on examples, including general chat turns so it still talks like a model, then merged into open weights.
Trained on launches before October 10, tested after. Scored against what really happened, and against Claude Opus 5.5 and Sonnet 5.5.
Two Charge models are live. On launches they never saw, both beat Claude Opus 5.5 at the question trenchers ask all day: will this double before it halves? Both exams are public. Next: Charge for pump.fun, and plain on-chain facts next to every read. Models ship only after a public exam. Facts ship with their source, never as a "safe" badge. Next means being built. Later means it waits until usage and fees pay for it. No dates we can't keep.
Paste any Pons launch here or in the bot: its charge, how launches like it did over the next 24 hours, and what stands out.
Robinhood Chain · Pons · free
Pons: 600 snapshots of launches it never saw, Charge 30B 0.91 vs Claude Opus 5.5 0.87. Direct: 1,000 launches it never saw, Charge Direct 0.90 vs Opus 0.65. Every answer, every outcome and the scoring scripts are on Hugging Face: Pons, Direct.
Hugging Face · rerun it yourself
Coins launched straight into Uniswap pools (plus Poolstrade and Bankr), rug rate included: 0.90 vs Opus 0.65 on 1,000 launches it never saw. You voted for it on day 1. It shipped on day 2.
Direct launches · 500k+ or burn credits
Link by signing a message (free, not a transaction), or from Fomo with a tiny burn. 500k+ unlocks Direct, 10x limits and your own API key. 2.5M+ adds new models first. $1 burned = 200 reads, and every burn is on the public ledger. See the tiers.
CHARGE · 500k+ and 2.5M+
Open weights on Hugging Face (Apache 2.0). Charge 30B on the Surplus API at $0.10 per million tokens, paid in CREDIT. A personal API key for Charged holders.
Hugging Face · Surplus API · API key
@ChargeAIRHBot reads any CA in DMs or groups (/read 0x…). Every read posts as a share card, and reading a fake CHARGE gets you the real CA.
Telegram · X
The biggest trench, and your second pick in the vote. Fresh pump.fun data with outcomes measured in SOL, then the same exam against Claude before it ships. Same token, and Overcharged holders get it first.
Solana · pump.fun · next up
Examined on the same 1,000 launches: 0.905 vs v1's 0.90 (Opus 0.65), and 0.72 vs 0.70 on launches it reads mid or hotter. A small step, not a fix: on quiet launches it still says about 6% rug when about 10% do.
Direct launches · Overcharged first
Charge's reads next to what actually happened, hits and misses both. Each receipt has a fee line: what came in and what went to GPUs, Surplus CREDIT, training and buybacks, with tx links.
Public · every model
The site and the bot move off a GPU we rent by the hour onto the Surplus API, paid in CREDIT from our fee share. That's what keeps Pons reads free.
Free · Surplus API
The model stays the same. The card around it gets more facts from the chain, one check at a time:
Every fact shows its source and when it was read. If we can't tell, it says unknown, never 0. No red/green verdicts and no "safe" badge. These sit next to the read. Anything that ever feeds a model takes the exam first.
Robinhood Chain · Pons and direct launches
Heavier on-chain checks that take more compute, more data and more care to get right in public. They come one at a time, as fees allow, and each is named when it ships.
as fees allow
Use Charge without holding CHARGE. A fixed, published share of every payment buys back and burns CHARGE, and each burn goes on the ledger with its tx.
no holding needed
For launches hours to weeks old that start trading again: is the wakeup real, or is one wallet pushing a dead chart?
Robinhood Chain
A proper API page for your own bots and agents, with batch reads for holder keys.
builders
Pons, direct launches and pump.fun in one model, trained on fresher tape, under the same token. On Surplus if they allow a newer model under a launched token, or on our own GPUs.
every trench
The order inside Next can change. This list won't.
Charge 30B is live on the Surplus API as : $0.10 per 1M tokens, paid in CREDIT. Keys, credit and the exact details are on Surplus.
No. Charge reads public trading facts and tells you how launches like this one did next. On Pons, even its hottest band halves first more often than it doubles; on direct launches, rugs are common in every band. What you do with a read is your call.
No. It reads the facts you give it: age, market cap and ATH, trades, buyers and sellers per window, holders, the dev's own trades, first-30-second buyers, the creator's other launches, socials. The Charge bot fills those in from a contract address.
Because the hottest tapes are the most crowded. Busy launches move in both directions: the runners come from the top bands, and so do most of the round-trips. A high charge says "this is where the action is", not "this is safe".
It doesn't read contracts, so honeypots and hook traps need a sell simulation. It can't see the team, the news or who is about to shill. It was tested on two days of one launchpad. It will be wrong, often.
$CHARGE launched on the Surplus LLM launchpad. Its creator fees go to a fixed splitter: 60% to the hosting vault that keeps Charge live on the Surplus API, 30% to the builder, 10% buys and burns $SURPLUS. The builder's 30% runs Charge: free reads on this site, in the Telegram bot and the API, GPUs before graduation, Surplus CREDIT after, training new models, and CHARGE buybacks from the public dev wallet, each announced with its tx.
$CHARGE unlocks Charge. Pons reads stay free for everyone. Link a wallet holding the Charged amount and you get Charge Direct (reads of direct launches), higher limits and a personal API key; the Overcharged amount adds first access to every new Charge model. Burn CHARGE for read credits when you need more. Every Charge model (Direct now, pump.fun next) ships under the same token. See the tiers.
500k and 2.5M are the launch tiers. If you qualify, we won't raise the bar on you: a price run never takes your tier away. As the price rises we lower the thresholds, so joining costs about the same in dollars. What each tier unlocks can grow (new models first), and read limits may be tuned to keep Charge running; heavy use is what burn credits are for.
The Charge read for coins launched straight into Uniswap pools on Robinhood Chain (plus Poolstrade and Bankr), where launches can also be rugged by pulling liquidity, so its reads include the rug rate. It's the first model for CHARGE holders. On 1,000 launches from October 13-14 it never saw, it scored 0.90 AUC at calling which ones double before they halve, vs 0.65 for Claude Opus 5.5 given the same facts (0.70 vs 0.51 on the launches it reads mid or hotter, the ones traders actually look at). It's a preview, so you get it early, limits included: it learned from October 5-11 only, it understates how often quiet launches halve or get rugged, it doesn't read contracts (honeypots, hook traps), socials or the team, it reads launches up to 3 days old, and it's weak at general crypto questions. Version 2 passed the same exam with a small gain; Overcharged holders get it first; its exam is public.
You sign a plain-text message: it's not a transaction, it can't move anything, and Charge never asks for approvals or allowances. Charge only reads your CHARGE balance and the burns your wallet makes. Burns are sent by you, from your own wallet, and are permanent.
If people use it, yes. The Surplus API serves the exact commit $CHARGE launched with, and Charge 30B's weights are on Hugging Face. Holder models like Charge Direct are served through chargeai.bid, the bot and the Charge API, examined the same way before they ship. Tell us on X what to train next.
Yes: t.me/chargeairh on Telegram. The Charge bot is in there too, so /read 0x… works in the chat. Admins never DM first.
The launch wallet can't buy on Surplus launches. Any dev buy is made from a public wallet, listed here.