Łapcie ciekawą stronę, która może być alternatywą dla Areny.
Ta strona to ASL, czyli AI Stupid Level.
Znajdziesz tutaj między innymi rankingi modeli w różnych kategoriach, testowane dziennie lub co 4 godziny.
Dodatkowo każdy model ma swoją stronę i możesz tam śledzić jego poczynania.
Są też inne opcje, takie jak np. "Drift", która pomaga śledzić czy dany model nagle nie stracił swoich możliwości, bo takie coś już się zdarzało. Tutaj niestety jest to wyłączone w opcji z jedyną uczciwą ceną.
Jest też Smart Router, czyli API, które automatycznie wybiera model odpowiedni do zadania, biorąc pod uwagę wyniki testów, szybkość i cenę.
Poniżej link:
aistupidlevel.info
Exciting news: Beam-1.0-501B-A23B by @Reflection_AI just landed #12 among open weight models in Agent Arena with -3.67% Net Improvement, and #22 among open in Code Arena: WebDev with 1438 pts. It’s also fully open under Apache 2.0.
On the Agent Arena leaderboard, Beam 1.0 is the top-ranked US model among open models, landing at #12. That puts it ahead of Hy3, MiMo V2.5 Pro and MiniMax M3. Its margin of error is wide (±3.84% on Net Improvement), so it’s statistically closer to the two Qwen3.8 models just above it than its rank suggests.
On Code Arena: WebDev, 24 of the top 25 open models come from Chinese labs. The one exception is Reflection’s Beam 1.0 (Apache 2.0). It ranks #22 among open models, which makes it the top US open model on the board, just ahead of Kimi K2.5 Thinking and GLM-5.
Congrats to the @Reflection_AI team on this solid open weights release!
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
nobrainer-tech retweeted
basal-omni 🇵🇱 - TOP1 🤩🤩🤩 🌍 na świecie w modelach otwartych oraz TOP2 wśród zamkniętych i otwartych
Wg niezależnego benchmarku modeli system-one gdzie testowane są wszystkjie modele na świecie - JevBench (link do bencha w pierwszym wpisie). Czyli można? Można! Polskie AI 🇵🇱
basal wyprzedził model OpenAI, Microsoft oraz pierwowzór z TypeSafe (jev).
Thank you, @airesearch12 , for the independent evaluation on JevBench.
nobrainer-tech retweeted
The 6 levels of building products with AI:
(From a ChatGPT prompt to a production app)
@AnkythShukla of HelloPM sorts every way of building with AI into 6 levels, so you can stop debating tools.
Here's what each level does, and when to climb:
L0. Prompting
ChatGPT, Claude or Gemini, straight out of the box. Good for a one-off question.
Move up when: you find yourself typing that same prompt every week.
L1. Reusable prompts
Save the prompt you keep retyping as a Google Gem or a custom GPT. A PRD Gem with your templates. A resume Gem for any job description.
Move up when: every task has you switching windows, and longer instructions make the output worse.
L2. Skills and connectors
Claude Code or Codex. The real return begins at this level.
A skill holds instructions for a single task and loads only when the chat needs it. Ankit's trick: finish the task in Claude, then have Claude make a skill out of it. Hook up your analytics for a 10am product brief.
Move up when: users or sales need an interface they can use without logging into Claude.
L3. Vibe coding
Ankit picks Google AI Studio, which is free. Lovable and Bolt work too.
Move up when: the work needs to run in the background without anyone opening a screen.
L4. Workflow tools
n8n or Make. A lead lands on your site, gets researched and scored, and reaches the right salesperson. The built-in AI assistant drafts the workflow from a prompt.
Move up when: you need production grade software on your own stack with full control.
L5. Production software
Cursor, MCPs into AWS or Vercel, Claude Code or Codex. In one night, Ankit replaced a $550 to $700 a month email platform here. It runs for $110 a month.
The rule: use the lowest level that gets the job done, and climb only once it breaks. L2 is enough for most PM work.
And quit debating models. Ankit's team built the same four sites with Claude Code, Codex, Cursor and DeepSeek. All four shipped.
What level are you at right now?
nobrainer-tech retweeted
GPT-6 Sol and Luna are out. Not only are they a very significant improvement across the board, but also in writing and general "you know when you try it" quality.
We are also permanently reducing the API price by 50% making both of them viable for a ton of new usecases and making your usage go further too, even on the subscriptions.
And one more thing. We are loading a banked reset into all accounts of our Plus, Pro and Business users. Let's go!
openai.com/index/introducing…
Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe.
GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale.
We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.
nobrainer-tech retweeted
Au lieu de regarder 2 heures de Netflix ce soir, regarde cette masterclass de 40 minutes du fondateur d'une entreprise chinoise d'IA valorisée à 20 milliards de dollars.
J'ai repris les idées essentielles et j'ai écris un thread 🧵
( AJOUTE EN SIGNET 🔖) :
C'est l'explication la plus claire que j'aie vue sur le fonctionnement réel des essaims d'agents (Agent Swarms) et des systèmes d'IA à grande échelle.
Que tu n'aies jamais créé un seul agent ou que tu utilises Claude tous les jours depuis un an, tu y trouveras quelque chose d'utile.
La suite de la vidéo est en bas ⤵️
nobrainer-tech retweeted
using OpenClaw?
setup these 3 cron jobs today:
1. session cleanup (every 72 hours)
→ deletes bloated session files that slow down your agent
2. daily security audit (every morning)
→ checks firewall, fail2ban, SSH, open ports, docker status
3. silent backups (every 2 hours)
→ git push your workspace so you never lose config/memory
nobrainer-tech retweeted
OpenAI is Apple 2026 (boring and cold)
Anthropic is Apple 1997 (creative and warm)
OpenClaw 2026.2.1 🦞
🔒 Major security hardening: path traversal, LFI, exec injection fixes
🧵 Discord thread routing + gateway message timestamps
🔐 TLS 1.3 minimum, system prompt guardrails
🛠️ Streaming stability, memory search fixes
20+ community PRs 🙏 github.com/openclaw/openclaw…
🦞 BIG NEWS: We've molted!
Clawdbot → Moltbot
Clawd → Molty
Same lobster soul, new shell. Anthropic asked us to change our name (trademark stuff), and honestly? "Molt" fits perfectly - it's what lobsters do to grow.
New handle: @openclaw
Same mission: AI that actually does things.
nobrainer-tech retweeted
We're moving from "AI assists" to "AI acts."
The people learning to work with autonomous agents NOW are building muscle memory for the future of work.
It's like learning spreadsheets in 1985 or search engines in 1998.
Early adopters aren't just saving time today.
They're developing fluency in a skill that will be mandatory in 5 years.
But here's the honest truth:
Most people won't invest the time to learn this properly.
They'll try it once, get frustrated when it doesn't instantly solve everything, and quit.
The real advantage goes to people who:
• Start with simple use cases
• Invest time learning what's possible
• Iterate and refine workflows
• Stay consistent
That's the group that will 10x their productivity.
nobrainer-tech retweeted
github.com/parcadei/Continuo…
continuous claude v2 is now up - a setup designed to tackle the scarcest resource in coding: context
explaining the reasoning behind features below ↓
nobrainer-tech retweeted
MCP servers were stuck on text and data. Not anymore. 👀
Proposed by Anthropic, OpenAI, and the MCP-UI community, the new MCP Apps Extension standardizes interactive interfaces with security built in.
Here's what you need to know. ▶️
nobrainer-tech retweeted
Who will win? Multi-billion dollar ai research lab or man with a markdown file?
My browser skill was over 2x faster at completing this task than Claude Code's new chrome integration!
Replying to @claudeai
Using the extension, Claude Code can test code directly in the browser to validate its work. Claude can also see client-side errors via console logs.
Try it out by running /chrome in the latest version of Claude Code.
nobrainer-tech retweeted
You can now type "&" in the Claude Code CLI to start a task in the background and teleport it to Claude Code Web!
nobrainer-tech retweeted
🔍🤖 Enterprise Deep Research
A multi-agent system leveraging LangGraph to power enterprise research automation. Features real-time streaming, human-guided steering, and flexible deployment through web and Slack interfaces.
Explore EDR on GitHub 📚
github.com/SalesforceAIResea…






























