230 Photos and videos
Following Anthropic's announcement, we are temporarily suspending access to Claude Fable 5 on the Creao platform until further notice.
All other Claude models remain fully available. We're monitoring the situation closely and will restore access as soon as Anthropic confirms compliance is resolved.
We apologize for any disruption to your workflows.
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
6
2
15
1,389
Job searching should not feel like a full-time job.
Students are checking LinkedIn, Handshake, company career pages, spreadsheets, deadlines, and email alerts — all while trying to study, build projects, and prepare for interviews.
This is exactly where agents can help you.
The Internship Tracker Agent finds SWE internships, extracts the key details, saves them to Google Sheets, ranks the best matches, and sends weekly updates.
No more scattered tabs. No more forgotten deadlines. No more starting from zero every week.
You can finally focus on what matters.
5
5
14
2,684
Claude Fable 5. Claude Opus 4.8. Gemini 3.5 Flash. GPT-5.5. MiniMax M3.
Five new models. Now in CREAO.
Pick the right model for each workflow, switch mid-conversation as the task evolves, and keep every file, memory, and context connected in one place.
Your work shouldn’t be scattered across five tools. It should flow through one system.
4
5
13
2,902
The whole world was watching Mythos. Waiting for it to drop.
It just did — and it's live on CREAO.
The headline: it runs complex tasks for days without losing focus. No drift. No re-engagement. First-shot accuracy, multi-agent coordination, and better visual accuracy that actually reads interfaces correctly.
Claude Fable 5, Available now.
4
4
16
6,980
The numbers back it up. And we saw it firsthand.
We've been running Fable 5 ahead of launch. The benchmark lead on long, complex tasks is real.
x.com/claudeai/status/206439…
Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision.
The longer and more complex the task, the larger Fable 5’s lead over our other models.
ALT Benchmark table titled Mythos 5 & Fable 5, comparing Claude Mythos 5 and Fable 5 against Claude Mythos Preview, Claude Opus 4.8, GPT 5.5, and Gemini 3.1 Pro.
3
573
CREAO just wrapped S1 of the Agent Trading Campaign in partnership with @WEEX_Official
The audit is complete and the final leaderboard is locked. 30 agents took home prizes from the $10K prize pool. The top performer turned 500 USDT into 3,182 USDT — a 536% return in one week with zero manual intervention. Nearly half of all agents finished in profit.
To all winners, please open a ticket on our Discord server to claim your prize.
7
5
16
837
The most interesting part of S1 wasn't who won, it was what the builders discovered along the way. As @furan86999 put it: "What's worth paying attention to isn't who placed first, but how these agents were created." So here's what we learned:
1/ ETH short dominated the top 10. The market rewarded one clear directional conviction over complex multi-signal setups. Simplicity won. @oldman_cek noted that roughly 90% of the top 50 ran some variation of ETH short. And the results confirmed it.
2/ But the most interesting story came from a participant who lost money. @Guomin184935 did four rounds of iteration: adding drawdown limits, Sharpe ratio protection, position sizing rules. Each version felt smarter. The result: the agent became so cautious it stopped trading. His reflection was that he kept optimizing for safety and forgot to optimize for returns. Over-engineering killed his edge.
3/ @0xDiyaaa had a similar experience: lost about 10%, then looked at the leaderboard and realized the top was all ETH short while they had stayed long. But they also noticed something important: nearly half of the agents finished profitable. The pipeline works. Strategy generation, risk parameters, autonomous execution, post-mortem, it all ran end to end.
4/ On the other side, @yupaopao0 finished #27 with just 6 trades. Built the entire strategy through conversation — no code. ETH short bias, 2-hour check intervals, technicals news sentiment macro events. Simple, disciplined, profitable.
5/ More trades did not mean more profit. Agent #7 made 943 USDT on 42 trades. Agent #8 made 897 USDT on 210 trades. Five times the activity, less return. Discipline beat frequency.
6/ None of these agents were built by quant teams. They were built by regular users describing their strategy in plain language, iterating through conversation, and then locking it in. As @cryptok64440829 observed: "With good tools, regular users can participate in Agent design and training — and even get great results." It's the beginning of the agent era for independent trading.
1
5
383
S1 proved something that matters beyond the leaderboard. As @jiamiguance put it: "The dividing line for AI trading products isn't whether they can discuss markets — it's whether they can trade for you."
Most AI trading tools today are what @yijiangren called "talking market commentators." They analyze, they summarize, they give opinions. But they don't execute. They don't manage risk over time. They don't run unsupervised for a week and come back with a P&L statement. S1 did exactly that.
The barrier between having a trading idea and executing it autonomously is disappearing. In the past, turning a market thesis into an automated strategy required code, APIs, infrastructure, and constant maintenance. Most people with real market intuition never got past the idea stage. S1 showed that's no longer the case. Users built, refined, and deployed trading agents through conversation alone.
But the deeper takeaway is this: agents don't just execute your strategy. They expose it. Every assumption, every bias, every blind spot gets surfaced when your agent runs unsupervised for 7 days in a live market. As @Guomin184935 reflected: "The boundary of the Agent is the boundary of the designer's cognition."
Trading stops being about speed or screen time. It becomes about clarity of thinking: who can articulate a strategy clearly enough for an agent to execute it, and who can resist the urge to over-engineer it.
S1 was 166 agents and a week of autonomous trading. It was also the first real stress test of this idea at scale. The results speak for themselves. S2 is coming.
3
289
Integration failures are one of the most common reasons agents break mid-run.
An agent runs fine for weeks. Then a connector token expires mid-run. The workflow stops. You find out later... sometimes much later, when the output never arrived.
Now when that happens, CREAO shows a Reconnect button directly in chat.
Click it, re-authorize, and the agent automatically retries from where it left off.
No context lost. No rebuilding the session. No starting over.
2
4
12
1,097
Your agent can build apps. Now it can make movies too.
CREAO is joining @billions_ntwk, plugging our agent harness into the network so partners and their customers get agents that actually run.
The world's first agentic movie with 1,024 fully AI-made tiles, onchain forever. And this is just the start.
Claim a tile and join the $100K leaderboard.
6
10
26
2,535
Automation that fails silently is worse than no automation at all.
If a scheduled run is skipped (low credits, something broke…) it now shows a Skipped badge and counts toward auto-pause.
Two consecutive skips and the schedule pauses automatically. You get an email explaining exactly why, with the steps to resume.
We'd rather interrupt you once than let a workflow quietly stop delivering for two weeks without you knowing.
14
3
20
1,489
Most agent frameworks assume a desktop: one user, one machine, one process.
Cloud agents have none of those luxuries — fresh sandboxes, shared hardware, callers you never meet, untrusted code.
Two lessons from rebuilding that layer at CREAO 🧵👇
3
2
16
2,046
Live from the S1 Trading Campaign:
78/165 participants are running profitable trading agents — and we're not even at the finish line yet.
Nearly half the field is already in the green.
The agent era of trading isn't coming. It's here.
Top 5 leading the pack: @karthikbuilder @saularis6 @Sahil349297 @hao_luo @zuoci1km
5
1
10
1,129
Your marketing team spends hours briefing, drafting, and chasing approvals every week.
This agent does it in 2 minutes.
Watch it:
→ Pull brand kit last quarter's top campaigns
→ Draft 5 on-brand ad variants for Q2
→ Drop them in #marketing
→ Schedule itself to do it again every Monday
5 ads. One campaign. Runs forever.
This is what "set it and forget it" actually looks like for marketing.
3
3
19
1,818
A Day zero is the only day that matters 🔥
The exciting part is the full stack together: coding frontier, 1M context, native multimodal input — running inside an agent harness where it can actually use all three.
Looking forward to seeing what people build with it!
@CreaoAI moving fast 🔥 M3's live on day one, go try it
6
12
1,127
M3 will stretch people's imagination. Sparse attention open-source = massive leap for AI builders.
x.com/MiniMax_AI/status/2061…
Introducing MiniMax M3: The First Open-Weights Model to Combine Three Frontier Capabilities
- Coding & Agentic Frontier: 59.0% SWE-Bench Pro, 66.0% Terminal Bench 2.1, 34.8% SWE-fficiency, 28.8% KernelBench Hard, 74.2% MCP Atlas
- MiniMax Sparse Attention scales context to 1M
- Natively Multimodal from Step Zero
API: platform.minimax.io
Token Plan: platform.minimax.io/subscrib…
🚀New! MiniMax Code: code.minimax.io
Weights & Tech Report in ~10 Days
5
9
936
Our co-founder @intuitiveml spent 6 years scaling LLaMA at Meta before co-founding CREAO.
There, he realized something: models got smarter, but teams still had to turn every answer into action manually.
He sat down with @unite_ai to explain why the next era of AI isn't a chat window but a closed-loop system that can connect to your work, audit what’s happening, build agents, run them, and improve over time.
Check the interview to understand why CREAO is building the missing loop👇
May 30
We rebuilt our entire company development and operation around AI. AI builds the tools, runs them, and improves them.
Now we're building a platform so any team can build the same self-improvement loop.
I did an interview talking through some of the details
• “AI-first” means redesigning the workflow, not adding AI to the old one.
• Engineers are moving from writing code to designing the systems that govern how agents write code.
• Self-healing software is closer than most people think.
• The real product is not the agent. It is the harness around the agent.
here's the link:
unite.ai/peter-pang-co-found…
5
17
36
1,900