Ambassador @DevinAI @cognition building @Sen_computer 👇Best way to get creative websites with SIMPLE prompts: 👇 https://teapaw.org/t.co/OTVkQuEyHs
Germany
Joined October 2024
- Tweets4.6K
- Following431
- Followers1.2K
- Likes8K
Pinned Tweet
GPT-6.2 SOL HIDING BEHIND GPT-6.1 SOL IN CODEX? 👀
I'm starting to believe it.
I tested GPT-6.1 Sol in Codex and via the API.
Same prompt. Same reference image.
Codex did a significantly better job.
The result was actually really close to the reference.
Does that prove routing? No.
But the difference was big enough to make me suspicious.
Unfortunately, without a reference image, the frontend design is still bad.
Anthropic is still way way way ahead in my testing.
3D generation looks really good though.
BREAKING
Google Gemini 4 Carbon is already working via Vertex AI 🔥
release today is 100% confirmed now.
Lets see what this beast can do.
106
Fable 5.2 killed Gemini 4 Carbon before it was released to the public.
I hope google learns and bring back the experimental releases.
11
2,441
Polymarket was right once AGAIN :)
Claude Fable 5.2 is rolling out.
FABLE IS SPIKING ON POLYMARKET 👀
"Next Fable model released on Nov 9" just jumped from about 20% to 41% in a few hours.
That's real money betting it drops TODAY.
Either someone knows something…
or a lot of larpers are about to lose money tonight.
Lets see who drops first GPT-6.2 Sol / Astra or Claude Fable 5.5 / Opus 5.6
2
310
GROK 4.7 JUST GOT HUMBLED BY A $0.14 MODEL 💀
Meta's smallest model, Muse Ember, beat it on the Artificial Analysis Index.
Ember: 36 points, $0.14 per task.
Grok 4.7: 35 points, about $7 per task.
You're paying 50x more for a worse score.
We need Grok 4.8 ASAP @elonmusk
Meta's newest model Muse Ember scores 36 on the Artificial Analysis Intelligence Index, matching Claude Haiku 5.5 (max) at a third of the Cost per Task
Muse Ember is Meta's newest efficient-tier model line, with the same input and output token prices as Claude Haiku 5.5 and GPT-6 Luna
Key Takeaways:
➤ On the Intelligence Index vs. Cost per Task Pareto frontier at both max and xhigh effort. Muse Ember (max) is the cheapest model scoring 36 or higher, at $0.14 per task, and Muse Ember (xhigh) scores 34 for $0.12. Claude Haiku 5.5 (max) costs almost 3x more at $0.41 for the same score as Muse Ember (max). Muse Ember (max) is also about 4x cheaper than DeepSeek V4.1 Flash 1017 (max, 33, $0.60) and GLM-5.3-Flash (28, $0.58)
➤ 8 points behind Muse Wonder for about 1/50th of the Cost per Task. Muse Wonder (max) scores 44 at $6.95 per Intelligence Index task
➤ Stands out on AutomationBench-AA. Muse Ember (max) scores 68.1% on AutomationBench-AA, level with Muse Wonder (max), and beats Claude Haiku 5.5 (max) on GDP.pdf (22.8% vs 20.8%), but trails it on AA-Briefcase (1521 vs 1574 Elo)
➤ Mixed in agentic coding in our Intelligence Index evaluations. It beats Claude Haiku 5.5 (max) on Terminal-Bench 4.0 (40.9% vs 32.8%) but trails it on FrontierCode 2.0 (27.2% vs 34.9%)
➤ Ahead of similarly priced models when running in its own coding agent. In Muse Code, Muse Ember (max) scores 46 on the Artificial Analysis Coding Agent Index, ahead of Claude Haiku 5.5 in Claude Code (max, 37) and GPT-6 Luna in Codex (max, 31)
➤ Knowledge is the main weakness, but it rarely hallucinates. Its AA-Omniscience accuracy of 21.1% is the lowest of the similarly intelligent models we compared, against 34.7% for Claude Haiku 5.5 (max) and 50.5% for Gemini 3.8 Flash (high). This is mostly because it declines to answer 64% of questions, which keeps its hallucination rate at 17.4%, half of Claude Haiku 5.5’s (max)
Other details:
➤ Pricing: $0.10 input / $0.004 cached input / $0.50 output per 1M tokens
➤ Reasoning effort: results shown at max effort, with xhigh also on the Intelligence vs. Cost chart ➤ Availability: Meta's first-party API and Muse Code
170
META'S SMALLEST MODEL IS BEATING CLAUDE HAIKU 🔥
Meta just dropped Muse Ember. It's their tiniest model, built for agent work.
On Meta's own chart it beats Haiku 5.5 and GPT-6 Luna:
- AutomationBench: 37.8 vs 13.9 and 20.2
- Coding Agent Index: 46 vs 37 and 31
- Terminal-Bench 4.0: 40.9 vs 39.2 and 16.4
1M context.
Frontend looks good for a fast small cheap model.
Below a oneshot website with the workofart skill.
FABLE IS SPIKING ON POLYMARKET 👀
"Next Fable model released on Nov 9" just jumped from about 20% to 41% in a few hours.
That's real money betting it drops TODAY.
Either someone knows something…
or a lot of larpers are about to lose money tonight.
Lets see who drops first GPT-6.2 Sol / Astra or Claude Fable 5.5 / Opus 5.6
DON'T SLEEP ON SWE-2 LIGHTNING ⚡
This website took 65 seconds to build.
Devin's SWE-2 Lightning built it using workofart, my free, open-source design skill.
65 seconds.
Fast models are getting scary good.
1
79
GPT-6.2 Sol incoming?
GPT-6.1 Sol on MAX effort on chatGPT work gives me other results compared to GPT-6.1 Sol via API.
Maybe some users are already getting routed to GPT-6.2 Sol
First impression: its slow but seems to be very capable.
Its testing time.
worked 23 minutes for this AK47 3D model
GPT 6.1 Sol is routing to another model, maybe its GPT 6.2 Sol (for me)
> SVG of BMW M4 CS side view
its not 6.1 sol, its easily better than that
3
10
3,547
Kimi K3.1 inside DevinCLI
Its really fast and reliable.
Here is a one-shot with the current workofart skill
4
268
COGNITION JUST DROPPED A NEW RELEASE 🔥
Devin Review 2.0 is out.
5x cheaper. And still faster.
If you're running multiple agents, lean hard into pull requests.
That workflow gets crazy productive.
With Devin Review 2.0 it's also safer.
You have no idea how many bugs I've already found just letting Devin review my PRs.
Yesterday SWE-2 flash
today Devin Review 2.0
what is coming next?👀
Introducing Devin Review 2.0: the best way to review PRs with Devin.
Now up to 5× cheaper, 1.5× faster, and with new features and integrations.
1
247
If this is true gpt 6.2 has to be a miss currently
OpenAI is in NO position to wait longer
Opus 5.5 is ahead in every aspect except computer use.
6.1 Terra is really really a great model and highly underrated
But for everything else OpenAI is so much behind.
Release the kraken @thsottiaux
You can hate or love the google cloud
But it never misses a leak. I love it 😂
Google cloud once again leaked new model
GPT-6.2 Astra and 6.2 Sol are coming probably tomorrow.
Today we got pretty randomly GPT-2.6 image
We might see fable 5.5 va Astra 6.2 tomorrow.
Fable is ready Anthropic is just waiting for the release.
Opus 5.5 is still after 6 weeks the king of all models so Anthropic has no pressure to release fable 5.5
i released workofart skill FREE and opensource on Github!
Here is a super basic 1 line prompt result with Sonnet 5.5 on high effort.
Would love any feedback.
INTRODUCING WORKOFART 🎨
My new skill for Devin, Claude, Codex & any AI agent.
It turns generic AI websites into sites that feel alive.
No more AI slop.
pushes the creativity to the limit.
IMO it works the best with Sonnet 5.5
And this is just v1.
I've got something really cool in the pipeline 👀
Free & open source.
1
84
INTRODUCING WORKOFART 🎨
My new skill for Devin, Claude, Codex & any AI agent.
It turns generic AI websites into sites that feel alive.
No more AI slop.
pushes the creativity to the limit.
IMO it works the best with Sonnet 5.5
And this is just v1.
I've got something really cool in the pipeline 👀
Free & open source.
MYTHOS 5.5 IS COMING SOON
can confirm. Chubby is right on the official website you can find Mythos 5.5 already in the code.
Next week we should get Fable 5.5 no doubts.
HUGE week!
Fable 5.5 / GPT 6.2 Astra / Qwen 4 and maybe GLM 5.5
i dont think we get Gemini 4 Carbon next week but proof me wrong google.
Mythos 5.5 is already listed HTML source code of Anthropic's official website
Thanks to the anonymous person who sent me the tip via email :)
1
2
166