Sharing the logic behind high-performance systems. Building projects on Solana.
Joined April 2024
- Tweets3.7K
- Following87
- Followers190
- Likes623
Pinned Tweet
Curiosity drives human progress.
From the wheel to quantum computing, every breakthrough began with someone asking "Why?" and "How?"
I'm driven by curiosity to understand:
- How modern machines think
- How quantum particles exist in superposition
- The nature of consciousness
- The mysteries of the cosmos
If you're curious too, about technology, science, philosophy, or the unknown, let's connect and explore together.
Because the future belongs to those who never stop asking questions.
now i am literally confused what to do from here. i am having 2 to 3 things at the same time i should do.
btw making a CEX
I just earned a certificate from @Microsoft . You can earn too from here:
learn.microsoft.com/api/cred…
solana is quietly getting faster every single week 👀
mainnet slot times just dropped from 400ms to 350ms
testnet is already testing 200ms
this is happening in the background while everyone's still arguing about TPS charts
what shipped this week:
- slot times cut again (testnet/devnet/mainnet all moving)
- Agave v4.3.0-beta.0 + v4.2.1 live
- Firedancer mainnet v0.1106.40201
- Token 2022 SDK, Program Metadata SDK bumped
- V1 transactions landing across Web3.js, Solana Kit, Solana Go
not marketing. not a roadmap post. just weekly shipping.
solana engineering doesn't sleep
🤖 Made with AI
do you want $20 of free api credits with zero credit card 🤑
stark relay gives new users ~$20 in credits
only email needed
what you get for fully free:
- $20 in api credits
- access to frontier models
- one endpoint, one key
- openai compatible
full setup guide (free):
step 1: go to starkrelay.bond
> open the site
> find the signup
step 2: sign up with email
> any email works
> no credit card
step 3: grab your key
> check the dashboard
> copy your api key
step 4: plug it in
> works in cursor, cline, aider, claude code
> one endpoint for the models
step 5: test one call
> send a prompt
> confirm it works
$20 of frontier api credits for $0
important: use it sincerely as it is free credits and not unlimited
this is a solid way to test frontier models free
bookmark this before it expires
don't waste $1000 for paid api. i found a github repo for your for FREE
public-apis is the biggest curated list of free apis on the internet
what you get:
- hundreds of free public apis
- 30+ categories: music, news, finance, weather, games, science, sports
- many need no auth or api key
- community maintained
- all free
categories:
- animals, anime, books
- business, calendar
- finance, games
- music, news
- science, sports, weather
- and way more
how to use:
1. open the repo
2. pick a category
3. grab the api you need
4. build something
no subscriptions, no paid tiers
just free apis for your projects
save this before you lose it
2 sites giving unlimited deepseek v4 flash free right now 😳
the model topping agent benchmarks and beats fable 5 on terminal coding and you dont need to pay a cent to use it
site 1: cnb.cool (unlimited till dec 31)
- deepseek v4 flash unlimited
- free until december 31, 2026
- no card, no subs
how to use:
1. open cnb.cool
2. open any repo
3. type
@codebuddy
4. turn on "work for me"
site 2: freebuff
- 5 hours of deepseek v4 flash free every day
- cli coding agent, open source
- beats most paid coding agents
how to use:
1. pnpm i -g freebuff
2. run freebuff in terminal
3. start coding with v4 flash
one model two free paths
your buddy pays for a coding agent
you get v4 flash on both for $0
bookmark this and use it whole year for free
now you can get around 1M api tokens daily on open models for free 🆓
no credit card needed
what you get for free:
- ~1M tokens per day on top open models
- llama and other open weights
- fast inference (cerebras hardware)
- no card, no subs
full setup guide (cerebras - free):
step 1: go to cerebras.ai
> open the site
> find the api / free tier
step 2: sign up
> email works
> no credit card
step 3: grab your api key
> open the dashboard
> copy your key
step 4: plug it in
> openai-compatible endpoint
> works in cursor, cline, aider
step 5: test one call
> confirm it replies
a million tokens a day for $0
hard to beat
save this and share with friends who need
finaly after 6 days of back and forth i completed the arbitrage contract with all tests, including mock test and real time mainnet test (forked with surfpool)
thanks to @solana_devs for making surfpool. such a og thing.
🇨🇳 China is giving away its AI model KIMI K3 for FREE.
now I replaced my entire $200/mo American AI stack with Chinese models
took a weekend but saved 90% cost and noticed the difference.
here's what I use for each thing 👇
• thinking through hard problems → Kimi K3
2.8 trillion params, 1M context, thinking mode built in it's the #1 Chinese model right now and the weights are free
• writing code → Qwen 3.7 Max
scores 80.4% on SWE-bench, ran 35 hours straight optimizing kernels I just paste my task and it goes. no handholding.
• agent tasks / tool calling → GLM-5.2
handles multi-step stuff better than anything at this price 1M context, MIT license, beats Claude Code on security
• reading entire codebases → Kimi K2.6
1M context window, spawns 300 agents in parallel
I dump a whole repo in and it comes back like a senior dev
• bulk processing → MiMo-V2.5-Pro (Xiaomi)
#3 Chinese model, matches Opus 4.6 at 20x the speed
$0.10/M tokens. yes, Xiaomi makes AI now.
• making images → Wan 2.7 (Alibaba)
open source, runs on my own GPU, free tier
upgraded from 2.5 now does video editing and multi-image stories
• making videos → Kling 3.0 (Kuaishou)
rivals Sora 2 at 1/5 the price
10 sec clip = $1 vs $5. native audio. 4K.
was paying $200+ a month to American APIs
now I pay ~$20
same quality and 90% less cost
free access list in the reply ↓
where to get them for free (limit applied):
huggingface.co/moonshotai Kimi K3 weights (free download)
together.ai K3 hosted day-0
modal.com K3 serverless
portal.nousresearch.com Kimi, GLM, Qwen
build.nvidia.com Kimi, GLM, DeepSeek (40 rpm free)
dash.cloudflare.com GLM 5.2 free tier
cloud.siliconflow.cn DeepSeek, Qwen, GLM
openrouter.ai all of them
for more FREE ai related tutorial, follow - @EnmiloX
i asked for some guidance from @solana_devs, but they didn't help me so I started working on my own.
i know there are some issues, like only one multisig for one pubkey but i am just testing the first one.
i will be happy if you give me some recommendations.
you can also earn $10 each for free.
how to claim it:
- download Plasma One: plasmaone.onelink.me/P8qq/rv…
- sign up with email
- use code, check comment section. the app is invite only.
- complete KYC
- you'll start on the lite tier
- deposit $100 to unlock core tier for free (check the offer first, once it ends the free upgrade is gone)
- if you spend $100 your will get 3% cash back + $10 bonus.
you can also get free ChatGpt subscription. here is how:
- their core tier gives you $100/year rebate on ChatGPT Go subscriptions
- that covers the entire annual cost
- plus 5% cashback on all AI
what you actually get:
- $10 bonus on spend and for each refer
- $100/year ChatGPT Go rebate, basically free
- 5% cashback on AI tools
- 3% cashback on everything else
this is a limited time offer. the free core tier upgrade won't last
Disclaimer: this is not a promotional tweet.
Hey @solana_devs, I want to build a multisig vault, but first I need to learn about it.
Can you suggest some beginner friendly resources that I can learn from.
I want to create the multisig from scratch so that I can understand the core rather than using any packages.