Track these devs in the @flapdotsh BNB trenches:
1️⃣0x6af3f52369c39f6ca2d9a50799723feff7ff679e
• Total launches: 32
• Graduated: 27
• Graduation rate: 84%
• Best: 牛来, peaked ~$163M
2️⃣0xc95b035dd0a56baeb5cda921d27b9d2b2409b24b
• Total launches: 9
• Graduated: 6
• Graduation rate: 67%
• Best: 蝴蝶人生, peaked ~$43M
3️⃣0x2d222aab2806580ce0f687ab694cd42b339d16f1
• Total launches: 8
• Graduated: 5
• Graduation rate: 63%
4️⃣0xa0f27f43fc145ecf66b838aae4711d0566fd3d31
• Total launches: 2
• Graduated: 2
• Graduation rate: 100%
• Best: 天秤, peaked ~$26M
5️⃣0x85de3316eeca740ba049a8d9c9d68ab21cb282e5 (MARSCOIN dev)
• Total launches: 1 on BNB + 2 on Robinhood Chain
• Graduated: 3/3
• Graduation rate: 100%
• Best: MarsCoin, peaked ~$266M
These are BNB deployers with some of the best graduation rates.
Since flap is backed by YZi Labs, whose founder is CZ himself, there's a good chance flap runners get listed on Binance (NFA).
A lot of Flap tokens actually got listed on Binance Alpha in the last 12 months and 2 even got listed on Spot.
Worth keeping an eye on these devs, and especially this launchpad.
Govak retweeted
Govak retweeted
One Golden Ticket has been hidden inside a Meteora DLMM Pro chocolate bar.
The lucky finder will receive access to a $1M DAMM v2 liquidity position.
Only at Breakpoint London, exclusively at the @met_lparmy Studio Lounge.
Govak retweeted
should've hidden those funds in private mode
Today, we are taking one of those leaps that change how crypto functions.
Introducing Solflare Private Mode, powered by @helius.
> Encrypted balances & tokens
> Private payments & transactions
> All at Solana speed
You can now encrypt your money within Solflare.
Live now in your wallet 👇
Govak retweeted
Yup! I'm the CTO bro. I will cook hard on this just hold and let me do all the cooking.
Govak retweeted
This market is wild
Why trade alone when you can just follow the pros?
Sounds good, right? 👀
A brand new feature is coming
Govak retweeted
I am launching a memecoin today.
Retweet and drop your fomo usernames below, I’m sending people cash until launch so you can buy.
The CA will be dropped here: t.me/trenchwithtrav
Govak retweeted
The Solflare Community Hub is LIVE NOW
$150,000 in Solflare merch up for grabs.
→ Sign in with X
→ Post tagging @solflare to stack points
→ Claim tees, hoodies, and more
📍 At Breakpoint? Collect at our booth
🌍 Anywhere else? Free shipping, pick your tier by Nov 18
First come, first served
hub.solflare.com
when Anthropic CEO realize that 8gb uncensored model on your own gpu just did the work opus 5.5 refused
Here's 4 steps to run an uncensored AI on your own GPU for $0:
I run a local abliterated Gemma 4 for anything I don't want leaving my machine
By the end, you'll know how to:
- run a model which has no refuses
- plug it into Claude Code
- do it all for $0, no API key, no account
So, let's go step by step:
1. Install Ollama
curl -fsSL ollama.com/install.sh | sh
you want a recent version, older builds don't speak the Anthropic API and don't support Gemma 4
2. Pull the model
ollama pull igorls/gemma-4-12B-it-heretic-GGUF
the biggest mistake people make here is grabbing a random "uncensored" tag with no quality numbers
this one is abliterated with Heretic. refusals went 99/100 → 0/100 at a KL divergence of 0.028 from the original, so you keep almost all of the base model's capability
one thing: it's still Google's Gemma license underneath, fine for personal use, but read the terms before you build a product on it
3. Turn off thinking (the part every other guide skips)
Gemma 4 is a hybrid thinking model and the abliteration targets non-thinking mode
leave thinking on and the reasoning block eats your token budget and the answer gets cut off
ollama run igorls/gemma-4-12B-it-heretic-GGUF
/set nothink
if you're calling the API instead, add "think": false to the request
4. Plug it into Claude Code (OPTIONAL)
ollama launch claude --model igorls/gemma-4-12B-it-heretic-GGUF
or manually:
export ANTHROPIC_BASE_URL=http://localhost:11434
export ANTHROPIC_AUTH_TOKEN=ollama
claude --model igorls/gemma-4-12B-it-heretic-GGUF
two gotchas:
- bump the context first. Ollama defaults to 4k and Claude Code's system prompt alone is bigger than that, so the model never sees its tool instructions. restart the server with OLLAMA_CONTEXT_LENGTH=32768 ollama serve (setting it in your shell does nothing if Ollama is already running as a service). this needs more VRAM than the 8GB below
- Gemma 4 tool calling is still rough, you should expect some failed edits, imo for serious agentic work use qwen3.5 or glm-4.7-flash instead
What you need:
- 8-12GB VRAM for Q4 (12B)
- ~8GB of disk for Q4, ~13GB for Q8
- more of both if you want a big context
honest take:
- "abliterated" means refusals removed, not quality added. it's still a 12B
- fewer refusals ≠ smarter. for real coding work I still use cloud models
- you own the weights, your prompts never leave the box. that's the actual win
*IMPORTANT: don't use it for anything illegal or harmful. that's on you, not the model
now i'm collecting a list of every uncensored model i'm adding to my local stack and which tasks each one replaces for me
also planning a big "Uncensored 101" article, still figuring out all the angles
save it so you don't lose it
Best Uncensored AI models now :
Overall : Qwen3.8-27B Abliterated
Small GPU : Gemma 4 12B Abliterated
Fastest : Gemma 4 26B-A4B Abliterated
Agentic : Qwen3.6-35B-A3B Uncensored
Writing : Mistral Small 3.2 Abliterated
No surgery : Hermes 4.3-36B
Image : FLUX.1-dev
Video : Wan 2.2
Ceiling : GLM-5.1 Abliterated (754B, not for home)
Where to get them :
all LLMs → huggingface.co/huihui-ai (search the name, grab the GGUF)
Hermes → huggingface.co/NousResearch
FLUX / Wan → ComfyUI + the model's HF page
Setup :
ollama pull hf.co/<repo>:Q4_K_M
ollama run <model>
/set nothink
that's it.
no API key, no account, no card, just few clicks setup
more on how to setup below:
Here's 4 steps to run an uncensored AI on your own GPU for $0:
I run a local abliterated Gemma 4 for anything I don't want leaving my machine
By the end, you'll know how to:
- run a model which has no refuses
- plug it into Claude Code
- do it all for $0, no API key, no account
So, let's go step by step:
1. Install Ollama
curl -fsSL ollama.com/install.sh | sh
you want a recent version, older builds don't speak the Anthropic API and don't support Gemma 4
2. Pull the model
ollama pull igorls/gemma-4-12B-it-heretic-GGUF
the biggest mistake people make here is grabbing a random "uncensored" tag with no quality numbers
this one is abliterated with Heretic. refusals went 99/100 → 0/100 at a KL divergence of 0.028 from the original, so you keep almost all of the base model's capability
one thing: it's still Google's Gemma license underneath, fine for personal use, but read the terms before you build a product on it
3. Turn off thinking (the part every other guide skips)
Gemma 4 is a hybrid thinking model and the abliteration targets non-thinking mode
leave thinking on and the reasoning block eats your token budget and the answer gets cut off
ollama run igorls/gemma-4-12B-it-heretic-GGUF
/set nothink
if you're calling the API instead, add "think": false to the request
4. Plug it into Claude Code (OPTIONAL)
ollama launch claude --model igorls/gemma-4-12B-it-heretic-GGUF
or manually:
export ANTHROPIC_BASE_URL=http://localhost:11434
export ANTHROPIC_AUTH_TOKEN=ollama
claude --model igorls/gemma-4-12B-it-heretic-GGUF
two gotchas:
- bump the context first. Ollama defaults to 4k and Claude Code's system prompt alone is bigger than that, so the model never sees its tool instructions. restart the server with OLLAMA_CONTEXT_LENGTH=32768 ollama serve (setting it in your shell does nothing if Ollama is already running as a service). this needs more VRAM than the 8GB below
- Gemma 4 tool calling is still rough, you should expect some failed edits, imo for serious agentic work use qwen3.5 or glm-4.7-flash instead
What you need:
- 8-12GB VRAM for Q4 (12B)
- ~8GB of disk for Q4, ~13GB for Q8
- more of both if you want a big context
honest take:
- "abliterated" means refusals removed, not quality added. it's still a 12B
- fewer refusals ≠ smarter. for real coding work I still use cloud models
- you own the weights, your prompts never leave the box. that's the actual win
*IMPORTANT: don't use it for anything illegal or harmful. that's on you, not the model
now i'm collecting a list of every uncensored model i'm adding to my local stack and which tasks each one replaces for me
also planning a big "Uncensored 101" article, still figuring out all the angles
save it so you don't lose it
Govak retweeted
Hey @grok, concise, no spin.
AMERICA’S FLIGHTS GROUNDED.
FAA COMMUNICATIONS FAILURE.
L3Harris runs 4,400+ FAA network sites. In 2025, it partnered and gave access to a FOREIGN STATE-OWNED MILITARY AEROSPACE AND INTELLIGENCE TECHNOLOGY CONTRACTOR.
Which country owns it?
Country only.
Govak retweeted
BREAKING: Nvidia, $NVDA, CEO Jensen Huang has resigned as CEO, President, and member of the Board of Directors, effective immediately.
how to get paid as an x creator
i made a list of kol agencies, some even take small accounts
the play: apply with your best post > take briefs that fit > post > get paid on delivery
> web three kol networks
- blockhive
- go to: blockhive.io/apply/
- exchange, defi and l2 campaigns, 613+ kols, review in 48h
- kolsklub
- go to: kolsklub.com/creator.html
- no joining fee, no exclusivity, matched on niche + your pricing
- creator wire
- go to: creatorwire.xyz/creators
- set your own rate per post, paid in usdc
- cryptoinfluencers io
- go to: cryptoinfluencers.io/joinask…
- apply > verify > get matched, paid in crypto
- sharky web3
- go to: sharkyweb3[.]com (remove [])
- 550+ kols, every one scored, campaigns run in private tg
- growgami
- go to: growgami.com/creators
- vc-backed crypto + fintech campaigns, x and tiktok, review in 48h
- surgence labs
- go to: surgence.io/creator-onboardi…
- 3,000+ vetted creators, never paid up front, wants creators who drive signups + on-chain actions
- icoda
- go to: icoda.io/services/creator-pr…
- 2,000+ creators on x, yt, tg, kick, ig. rate + exclusivity shown before you accept
- coinbound
- go to: coinbound.io/apply/
- 800+ influencer roster, one of the oldest crypto agencies
> no follower floor, stated outright
- peeked
- go to: peeked.io/creators
- "under 10k counts. there is no minimum". no fee, no cut of your rate
- cry pto ugc
- go to: cryptougc.co/crea
tors/
- paid per accepted video every week, stablecoin option, face optional
> what not to do
- don't pay any agency a joining fee
- don't trust tg "agency" dms. apply on the site
- don't buy followers. agencies audit for bots before they pay
> the catch
- applying =! getting campaigns. they match by niche + impressions, not followers
- rates are per campaign. read the brief before you post
- paid posts need x's paid partnership label, respect it
advice: post daily in one niche for a month, then apply with your best post as the sample
I'm also searching for small creators, if you are one message me, will add you to my network
bookmark this & take actions
Govak retweeted
Uncovering the MAX scam:
I found a cluster of wallets that made over $400,000 in profits
- all fresh wallets
- bought token at same time frame
They sent profits later to the same wallet (6QEsn7DjRjzP48ujMxW89zEaijsEdrqLURgwUvNSCAfL)
From there they transferred everything to coinbase
Govak retweeted
15 Underhyped Crypto Tools You Need Right Now:
@alphagate_io - Early project discovery
@WhalesMarket - Pre-market token pricing
@InsightXnetwork - Insider wallet detection
@moniai_ - Smart account tracking
@sparsitydata - Holder conviction signals
@HolderScan - Real holder growth
@ICO_Analytics - TGE & launch performance
@_dexuai - Narrative & social trends
@revertfinance - LP profitability analysis
@RootDataCrypto - Funding & token unlocks
@TenderlyApp - Transaction simulations
@artemis - Chain fundamentals
@SurfAIHQ - AI-powered crypto research
@BlockSecTeam - Onchain transaction tracing
@CieloFinance - Smart wallet trade alerts
Most offer free tools or a free tier.
Bookmark this list to get an edge.
how to get x premium for free
mlh's global hack week is giving every participant a promo code for 1 free month of x premium
the play: register > check in > get the code by email > redeem on x
> signup
- go to: events.mlh.com/events/14729-…
- log in or sign up for mlh, then register for "global hack week: builders week"
- connect your github and verify your mobile number
> steps
- check in to the event
- open the challenge "sign up for an x premium account"
- wait for the mlh email with your promo code link
- no email? find the mlh coach in the event dc server and ask for a code
- redeem it on x, the checkout should show $0 for month 1
- screenshot your premium dashboard and submit it to the challenge for swag
> what you get
- x premium for 30 days
- per mlh's partner page, one premium sub covers cursor, grok bot and the grok api
> who it's for
- open worldwide: students, working professionals
- your mlh account needs to be 30+ days old, so new accounts might not get a code this round
> what not to do
- don't wait for the code after nov 12, the event closes
- don't pay anyone for a "code"
- don't forget to cancel the subscription. it auto renews after the free month
> the catch
- codes come by email and can take about a day
- reportedly they expire around the end of nov, so redeem when it lands
> next one?
- mlh's next global hack week is open source week, dec 4 - 10
- no premium offer announced for it yet. make your mlh + github accounts now so you clear the 30 days old account rule
advice: make an mlh account today even if you miss this one. you can join next campaigns
this campaign ends tomorrow
bookmark this & take actions before tomorrow night
how to make money doing arbitrage
there's a ~5% price gap between lighter & variational
the play: compare prices > match exposure > check costs > track the gap
> how it works
- long the cheaper openai perp
- short the more expensive one
- close both if the gap narrows enough to cover your costs
you're trading the difference between two prices, not simply predicting openai goes up
> first step
- check the actual buy quote on variational
- check the actual sell quote on lighter
- use your intended order size, not the price on a chart
- compare the contract specs before treating them as the same exposure
> size both sides properly
- match the underlying exposure after checking contract multipliers
- equal deposits don't automatically mean an equal hedge
- keep margin available on both venues
- profit on one venue can't automatically save a liquidation on the other
> the catch most people skip
- these are derivatives, not openai shares
- variational derives its pre ipo index from external venues
- lighter uses internal pricing before ipo
- variational says its contract transitions after ipo; lighter says its pre ipo market settles and delists
same ticker. different pricing and settlement rules. nothing guarantees the gap closes on your schedule
> check the actual costs
- funding on both legs
- entry and exit spreads
- trading fees and slippage
- moving collateral between venues
variational's published pre ipo funding is fixed at 0.005% every 8 hours. don't assume both legs cancel each other's funding
> what not to do
- don't call the displayed gap your profit
- don't add leverage just because you've opened a hedge
- don't count points as cash or a confirmed airdrop
- don't deposit before checking both venues' country restrictions
less exposure to market direction doesn't mean less exposure to every other risk
bookmark this & dyor, i can make detailed guide on farming funding fees, just comment
how to get web3 jobs
web3 jobs do not care about your degree, your experience and skills matter the most
the play: pick 1 role > build public proof > make friends > get seen by the people hiring
> what people think vs the reality
- "you need to code" > content, community, support, research and design are real roles too
- "just apply on job boards" > apply, but founders also notice people whose replies they already know
- "a resume is enough" > your x profile and pinned post are part of your resume now
- "post gm every day and you'll get noticed" > nobody hires the gm guy, they hire the person who explained their product better than they did
- "big projects pay best" > small teams often hire faster and let you do more
> the roles that actually hire
• content writer
- writes posts, articles, docs and announcements
- skills: explain a protocol in plain words, hooks, editing, separating live features from roadmap
- proof: 3 pieces on one live protocol (a tutorial, an announcement, a "who should use it" comparison)
• ghostwriter
- writes as a founder on their personal x
- skills: copying someone's voice, interviewing, hooks, staying confidential
- proof: turn a public founder interview into 3 posts, labeled as practice samples
• community manager
- runs dc and tg, handles faqs, mods, spaces
- skills: patience, fast replies, spotting impersonators, escalating real bugs
- proof: a sample faq + an escalation guide for a failed transaction
• support / technical support
- figures out if a problem is the wallet, the network, the bridge or the app
- skills: reading block explorers, approvals, gas, stuck transactions, clear bug reports
- proof: 3 troubleshooting guides (stuck approval, failed swap, delayed bridge)
• kol / bd manager
- finds kols, partners and listings, then closes deals
- skills: cold dms, negotiating, tracking deals in a sheet, knowing who's real and who's botted
- proof: a list of 20 real partners for a project, with why each one fits
• researcher / analyst
- writes project deep dives, token breakdowns, airdrop guides
- skills: reading docs, unlocks and emissions, dune basics, spotting red flags
- proof: compare 3 protocols with sources and what you couldn't verify
• designer / video editor
- banners, explainers, short clips
- skills: figma or canva, capcut, fast turnaround, working from a brief
- proof: a product explainer carousel using the real interface
> first steps
- pick one role, not five
- fix your x bio: role + niche, e.g. "web3 content writer | defi + ai agents"
- pin your best piece of work
- follow 30 founders and teams in the niche you want
> steps
- every day: 10 replies on founders' posts that add a number, a step or a catch
- every week: post 1 piece of proof
- after 2 weeks: apply with a short sample made for that team, not just a resume
- go to: web3[.]career and cryptojobslist[.]com for listings
> what not to do
- don't dm "gm sir any job?", send work instead
- don't pay anyone a "deposit" or "training fee" to unlock a job
- don't run repos or extensions a "recruiter" sends on the machine holding your wallets
- don't do weeks of unpaid work for "tokens later", ambassador roles aren't always paid jobs
> the catch
- a lot of web3 pay comes in tokens, ask for vesting terms and a stable cash part
- agree scope, pay dates and revisions in writing before you start
- it's slow at first, most people quit before week 3
one role, real proof, in front of the people hiring. that's the whole game
you need to make friends in the journey, you need good connection to get most of the the jobs
bookmark this before you send your next "gm sir any job" dm




































