Pinned Tweet
FREE AI MODELS. NO CREDITS. NO CARD. 🔥
Numes is offering free access to some powerful models:
DeepSeek V4.1 Flash
GLM 5.3
GLM 5.3 Flash
Kimi K3.1 Pro
Nemotron 3 Ultra
you also get an OpenAI-compatible API that works with coding agents.
they’ve recently increased the free model rate limits by 4x.
just sign up, create an API key and start experimenting.
try it: numes.ai
definitely worth checking out
kadd retweeted
Anthropic just published how they build agents that keep working while you sleep. Here is how it works.
The agent isn't always on. It sleeps most of the time and wakes up for short runs. Five pieces around the normal agent loop make that work.
1. Event inbox. Webhooks, schedules and user replies go into one database table instead of calling the agent directly. Only one worker reads it at a time, so two events can't start competing runs and a crash doesn't lose an event.
2. Timers. A separate table holds follow-ups the agent sets for itself and schedules the user sets. A scheduler checks it and drops due ones into the inbox.
3. Memory. A folder of files. instructions.md, index.md and reported.md load on every run. reported.md is what the agent already told you, so it doesn't repeat itself. Notes and one progress file per task load only when needed. Save only what the agent can't look up again.
4. Permissions. Credentials sit in a proxy outside the agent that adds them to outgoing requests. Risky actions become deferred tool calls. The run ends, the session is saved mid-call, and your answer comes back as a new event that resumes it.
5. Outbox. Every run ends one of three ways. It sends you a message, decides nothing is worth sending, or asks for your approval.
Their advice on messages is to start quiet and earn more of your attention over time, with evals to find the right amount.
kadd retweeted
CODEX CLI 0.171.0 IS HERE 👀
OpenAI just shipped a new update with some useful changes:
- GPT-6.2 Sol is now the default model
- GPT-6.2 Astra and Sol added to the model catalog
- TUI workspace is now enabled by default
- preview documents, images and -subagent transcripts directly in the terminal
- new undo and redo shortcuts
update with npm:
npm install -g @openai/[email protected]
release notes: github.com/openai/codex/rele…
NVIDIA’S 550B AI MODEL IS FREE ON OPENROUTER 🔥
Nemotron 3 Ultra is available for free, with:
1M token context window
550B total parameters / 55B active
strong reasoning and coding capabilities
built for AI agents and long-running workflows
free tier rate limits apply.
try it here: openrouter.ai/nvidia/nemotro…
worth testing if you want a powerful long-context model without paying for API access.
kadd retweeted
NVIDIA’S 550B AI MODEL IS FREE ON OPENROUTER 🔥
Nemotron 3 Ultra is available for free, with:
1M token context window
550B total parameters / 55B active
strong reasoning and coding capabilities
built for AI agents and long-running workflows
free tier rate limits apply.
try it here: openrouter.ai/nvidia/nemotro…
worth testing if you want a powerful long-context model without paying for API access.
kadd retweeted
FREE AI API ACCESS. NO CREDIT CARD NEEDED 🔥
Groq gives you free, rate-limited API access to open models like GPT-OSS, Qwen and Whisper.
just sign up, grab your API key and start building.
no payment details needed.
get started: aifree.dev/offer/groq-free
try it
CODEX CLI 0.171.0 IS HERE 👀
OpenAI just shipped a new update with some useful changes:
- GPT-6.2 Sol is now the default model
- GPT-6.2 Astra and Sol added to the model catalog
- TUI workspace is now enabled by default
- preview documents, images and -subagent transcripts directly in the terminal
- new undo and redo shortcuts
update with npm:
npm install -g @openai/[email protected]
release notes: github.com/openai/codex/rele…
FREE AI API ACCESS. NO CREDIT CARD NEEDED 🔥
Groq gives you free, rate-limited API access to open models like GPT-OSS, Qwen and Whisper.
just sign up, grab your API key and start building.
no payment details needed.
get started: aifree.dev/offer/groq-free
try it
CODEX CLI 0.171.0 IS HERE 👀
OpenAI just shipped a new update with some useful changes:
- GPT-6.2 Sol is now the default model
- GPT-6.2 Astra and Sol added to the model catalog
- TUI workspace is now enabled by default
- preview documents, images and -subagent transcripts directly in the terminal
- new undo and redo shortcuts
update with npm:
npm install -g @openai/[email protected]
release notes: github.com/openai/codex/rele…
kadd retweeted
Gemini 4 Argon and GPT-6.2 Sol are FREE to try
You can also try more ai models like:
- Gemini 4 Argon,
- GPT-6.2 Sol
- Claude Sonnet 5.5
- Kimi K3.1 Pro
- GLM-5.3
you can also use Genspark GenCode CLI for coding directly from your terminal.
Create account and just open AI Chat, select your preferred model, and start building.
and since it's limited, I think it's good for everyday tasks.
Try it here: genspark.ai/agents?type=ai_c…
kadd retweeted
BREAKING: WorkBuddy now lets you use multiple AI models in one workspace!
No more opening different platforms, managing multiple subscriptions, or jumping between tabs just to access different AI models.
WorkBuddy brings them together in one place. 🔥
Here's what makes this update worth checking out:
One workspace. Multiple AI models.
You can access different models within WorkBuddy and switch between them depending on the task you're working on.
Need help writing content? Choose a model.
Working on research or complex questions? Switch to another model.
Want to compare responses or approach a problem differently? Change models without leaving your workspace.
You don't have to rebuild your workflow every time you want to try a different AI model.
Simply select the model you want to use and continue working within the same workspace.
Here's what's available:
• Hy4 Flash is FREE until November 30.
• Users can claim FREE credits until November 30.
• The Pro plan costs just $7, with access to GPT-6-Astra, Gro-4.7, and Gemini-3.8-Flash.
Instead of being locked into a single model, you get the flexibility to choose the one that fits your needs.
Different models have different strengths. Having them accessible from one workspace makes experimenting, comparing outputs, and getting work done much easier.
Less platform switching. More flexibility. More ways to work with AI.
One workspace. Multiple models. Your choice.
Check out @WorkBuddy_AI.
🤝 Paid partnership
kadd retweeted
OPENAI IS GIVING 50,000 RESEARCHERS FREE CHATGPT PLUS
applications open november 23, and grants are first-come, first-served.
each grant includes:
12 months of free ChatGPT Plus
access for the recipient and 2 collaborators
up to 150,000 researchers covered in total
it’s aimed at eligible academic faculty and postdocs using AI for research.
if you qualify, get ready to apply when applications open.
details: openai.com/index/expanding-c…
kadd retweeted
STUDENT FREE AI OFFERS ARE STILL ONGOING, IN CASE YOU MISSED THEM🔥
here are some you can still claim:
Google AI Pro: 1 year free + 5TB storage for eligible U.S. students: one.google.com/ai-student
GitHub Copilot: free while verified as a student: education.github.com/pack
OpenAI Codex: $100 credits for eligible U.S. and Canadian students: chatgpt.com/codex/students
Kiro: 1,000 credits/month for a year: kiro.dev/students/
Zed Pro: free for eligible students: zed.dev/education
v0: free student access: v0.dev/students
JetBrains: free student developer pack: jetbrains.com/community/educ…
Azure: $100 free credit for students: azure.microsoft.com/en-us/fr…
some offers have country or university restrictions, so check your eligibility and claim what you can
don't sleep on these
OPENAI IS GIVING 50,000 RESEARCHERS FREE CHATGPT PLUS
applications open november 23, and grants are first-come, first-served.
each grant includes:
12 months of free ChatGPT Plus
access for the recipient and 2 collaborators
up to 150,000 researchers covered in total
it’s aimed at eligible academic faculty and postdocs using AI for research.
if you qualify, get ready to apply when applications open.
details: openai.com/index/expanding-c…
kadd retweeted
DEEPSEEK V4.1 FLASH IS STILL FREE. NO CREDIT CARD NEEDED 😳
you can access it for free through multiple platforms:
OpenCode (inside the client only)
NVIDIA NIM
Token Harbor
no need to pay for the official API just to try it out.
try them here:
OpenCode: opencode.ai/
NVIDIA: build.nvidia.com/deepseek-ai…
Token Harbor: tokenharbor.ai/
each platform has its own usage limits, but definitely worth checking out.
STUDENT FREE AI OFFERS ARE STILL ONGOING, IN CASE YOU MISSED THEM🔥
here are some you can still claim:
Google AI Pro: 1 year free + 5TB storage for eligible U.S. students: one.google.com/ai-student
GitHub Copilot: free while verified as a student: education.github.com/pack
OpenAI Codex: $100 credits for eligible U.S. and Canadian students: chatgpt.com/codex/students
Kiro: 1,000 credits/month for a year: kiro.dev/students/
Zed Pro: free for eligible students: zed.dev/education
v0: free student access: v0.dev/students
JetBrains: free student developer pack: jetbrains.com/community/educ…
Azure: $100 free credit for students: azure.microsoft.com/en-us/fr…
some offers have country or university restrictions, so check your eligibility and claim what you can
don't sleep on these
CODEX CLI 0.171.0 IS HERE 👀
OpenAI just shipped a new update with some useful changes:
- GPT-6.2 Sol is now the default model
- GPT-6.2 Astra and Sol added to the model catalog
- TUI workspace is now enabled by default
- preview documents, images and -subagent transcripts directly in the terminal
- new undo and redo shortcuts
update with npm:
npm install -g @openai/[email protected]
release notes: github.com/openai/codex/rele…
NVIDIA’S 550B AI MODEL IS FREE ON OPENROUTER 🔥
Nemotron 3 Ultra is available for free, with:
1M token context window
550B total parameters / 55B active
strong reasoning and coding capabilities
built for AI agents and long-running workflows
free tier rate limits apply.
try it here: openrouter.ai/nvidia/nemotro…
worth testing if you want a powerful long-context model without paying for API access.
kadd retweeted
10 FREE BRITISH AI VOICE GENERATIONS DAILY. NO CREDIT CARD REQUIRED 😳
British Accent Generator gives you:
315 UK voices (RP, Scottish, Welsh, Northern Irish)
10 free credits every day
daily credit refill
short voice generations for free
perfect for short clips, voiceovers and testing.
try it here: britishaccent.pro/
don’t sleep on this
DEEPSEEK V4.1 FLASH IS STILL FREE. NO CREDIT CARD NEEDED 😳
you can access it for free through multiple platforms:
OpenCode (inside the client only)
NVIDIA NIM
Token Harbor
no need to pay for the official API just to try it out.
try them here:
OpenCode: opencode.ai/
NVIDIA: build.nvidia.com/deepseek-ai…
Token Harbor: tokenharbor.ai/
each platform has its own usage limits, but definitely worth checking out.
NVIDIA’S 550B AI MODEL IS FREE ON OPENROUTER 🔥
Nemotron 3 Ultra is available for free, with:
1M token context window
550B total parameters / 55B active
strong reasoning and coding capabilities
built for AI agents and long-running workflows
free tier rate limits apply.
try it here: openrouter.ai/nvidia/nemotro…
worth testing if you want a powerful long-context model without paying for API access.
WAIT… TENCENT JUST DROPPED A 770B PARAMETER AI MODEL.
Hy4 is now available with open weights.
→ 1M-token context window
→ image and video understanding
→ available on Hugging Face
developers can now download and experiment with it.
open-weight AI is getting serious.
10 FREE BRITISH AI VOICE GENERATIONS DAILY. NO CREDIT CARD REQUIRED 😳
British Accent Generator gives you:
315 UK voices (RP, Scottish, Welsh, Northern Irish)
10 free credits every day
daily credit refill
short voice generations for free
perfect for short clips, voiceovers and testing.
try it here: britishaccent.pro/
don’t sleep on this
DEEPSEEK V4.1 FLASH IS STILL FREE. NO CREDIT CARD NEEDED 😳
you can access it for free through multiple platforms:
OpenCode (inside the client only)
NVIDIA NIM
Token Harbor
no need to pay for the official API just to try it out.
try them here:
OpenCode: opencode.ai/
NVIDIA: build.nvidia.com/deepseek-ai…
Token Harbor: tokenharbor.ai/
each platform has its own usage limits, but definitely worth checking out.



















