Opus 4.6 one shot your million dollar website
Based Dario Amodei https://t.co/CDoCPgtdHr
Based Dario Amodei https://t.co/CDoCPgtdHr
17
20
461
72.8K
244
Gemini 3.0 Pro GA is CRAZY. The SVG's I have seen today blew me away every time.
I know I said I was not going to hype Gemini, however, this feels too good to not share, if they nerf it I am going to be the first one to tell you about it
Opus 4.6 feels irrelevant already. I https://t.co/HvrNQriwml
I know I said I was not going to hype Gemini, however, this feels too good to not share, if they nerf it I am going to be the first one to tell you about it
Opus 4.6 feels irrelevant already. I https://t.co/HvrNQriwml
49
25
475
143.7K
116
Claude Opus 4.6 has been spotted.
This is separate from the misinterpretation of Sonnet 5 information that led people to definitively assert the release date was this Tuesday.
This is separate from the misinterpretation of Sonnet 5 information that led people to definitively assert the release date was this Tuesday.
40
23
687
259.3K
75
π¨Claude Sonnet 5
knowledge cut off - may 2025
this is the non thinking version output and it crushes the thinking version of all companies on same prompt , tomorrow is the big day
prompt and codepen link in thread https://t.co/I5u243k9yL
knowledge cut off - may 2025
this is the non thinking version output and it crushes the thinking version of all companies on same prompt , tomorrow is the big day
prompt and codepen link in thread https://t.co/I5u243k9yL
40
35
792
231.8K
341
π¨ Claude Code LEAKS
It can now spawn specialized agents that work on tasks like teammates
β Each gets a detailed brief and builds autonomously β Runs in background while you keep chatting
β Multiple agents work in parallel on different parts
Basically a dev team in your
It can now spawn specialized agents that work on tasks like teammates
β Each gets a detailed brief and builds autonomously β Runs in background while you keep chatting
β Multiple agents work in parallel on different parts
Basically a dev team in your
39
31
703
172.7K
196
Fennec π¦ will mogs Snowbunnyπ
> 1 million context
> 1/2 the price of opus 4.5 < better in all area>
> trained on TPUs
>Faster will mogs every model in agentic coding
model information from Vertex, Sonnet 5 is expected to be released as early as next week. https://t.co/E0B2t5XC2D
> 1 million context
> 1/2 the price of opus 4.5 < better in all area>
> trained on TPUs
>Faster will mogs every model in agentic coding
model information from Vertex, Sonnet 5 is expected to be released as early as next week. https://t.co/E0B2t5XC2D
62
49
1.0K
2.7M
185
claude-sonnet-5-20260203
79
88
2.2K
4.4M
229
I want to say it loud
This is better , cheap and faster than Opus 4.5 with 1 m context window
Fennec π¦ coming soon , and claude code is also getting update ( your agents will talk to each other )
Claude code will decimate the market and can't spill more tea β
This is better , cheap and faster than Opus 4.5 with 1 m context window
Fennec π¦ coming soon , and claude code is also getting update ( your agents will talk to each other )
Claude code will decimate the market and can't spill more tea β
63
33
1.0K
455.2K
259
Big week for Anthropic fans coming upπ
(Or perhaps just anyone who uses AI to code)
(Or perhaps just anyone who uses AI to code)
69
30
932
4.7M
85
Google now provides GCP credits in your AI Pro and Ultra subscriptions
AI Pro members get $10 monthly
AI Ultra members get $100 monthly https://t.co/YDuHzzQz26
AI Pro members get $10 monthly
AI Ultra members get $100 monthly https://t.co/YDuHzzQz26
99
101
1.7K
4.1M
770
I had access to this beast for the last 7 days
Many of you guys guessed it right,
Agents swarms - 100 subagents working in parallel, 1500 tools call
Sota on HLE - 50.2% and Browse Comp- 74.9% https://t.co/CvUTc0DHVt
Many of you guys guessed it right,
Agents swarms - 100 subagents working in parallel, 1500 tools call
Sota on HLE - 50.2% and Browse Comp- 74.9% https://t.co/CvUTc0DHVt
28
16
488
51.5K
140
Kimi k2.5 around a year has passed
same prompt and you can see the improvement
One shot and check quoted post for prompt and to see how far open source has come https://t.co/Yx0vc1MeKB
same prompt and you can see the improvement
One shot and check quoted post for prompt and to see how far open source has come https://t.co/Yx0vc1MeKB
24
16
579
257.4K
285
The models I know that will come out 100% in the next few weeks:
GPT 5.3 model update along with 5.3 Codex.
New Claude Sonnet update
New Gemini 3 Pro update, GA or even 3.5, my guess is GA.
I don't have high hopes for Gemini model, however, someone said that Google has
GPT 5.3 model update along with 5.3 Codex.
New Claude Sonnet update
New Gemini 3 Pro update, GA or even 3.5, my guess is GA.
I don't have high hopes for Gemini model, however, someone said that Google has
42
31
685
119.6K
84
Gpt-5.3 will be better in every aspect
( but they are cooking something good , for STEM - reasoning capabilities will lead to many novel solutions π)
GPT - 5.2 had issues during pre training that's why writing and creativity is worse than gpt-5.1 confirmed by @tszzl
( but they are cooking something good , for STEM - reasoning capabilities will lead to many novel solutions π)
GPT - 5.2 had issues during pre training that's why writing and creativity is worse than gpt-5.1 confirmed by @tszzl
29
23
515
80.6K
73
π¨ Gemini Snowbunny - ONE SHOTTED
Google really cooked here , prompt is in the quoted tweet , see the windows 11 clone
I dont want to hype it , like we never get these beautiful model in public, like FR compute is the only bottleneck for all labs https://t.co/GB550o7CFW
Google really cooked here , prompt is in the quoted tweet , see the windows 11 clone
I dont want to hype it , like we never get these beautiful model in public, like FR compute is the only bottleneck for all labs https://t.co/GB550o7CFW
40
43
642
96.7K
223
Two versions of an upcoming Gemini model - codename Snowbunny - both achieve a state-of-the-art performance on Heiroglyph, my benchmark testing lateral reasoning capabilities.
This model is in a league of its own for these tasks. Looks like I'm gonna have to start working on V2. https://t.co/6HozBynoaT
This model is in a league of its own for these tasks. Looks like I'm gonna have to start working on V2. https://t.co/6HozBynoaT
66
54
897
517.6K
224
Snowbunny - Gemini 3.5 early checkpoint or can be pro GA
> Our team is testing this on studio for a week at least , soon we will post various types of test
> From frontend to back end from 3d to music
> this model feels something big like deepthink but fast like flash https://t.co/yNCCxQNYZ5
> Our team is testing this on studio for a week at least , soon we will post various types of test
> From frontend to back end from 3d to music
> this model feels something big like deepthink but fast like flash https://t.co/yNCCxQNYZ5
25
26
467
59.0K
97
So google started A/B testing again in Ai studio :
ONE checkpoint are being tested ( maybe it's 3 Pro GA )
> It's far better - like they can say it's 3.5 and people will accept it
> We will start sharing results follow these : @marmaduke091 @legit_api @Lentils80 @MarsForTech
ONE checkpoint are being tested ( maybe it's 3 Pro GA )
> It's far better - like they can say it's 3.5 and people will accept it
> We will start sharing results follow these : @marmaduke091 @legit_api @Lentils80 @MarsForTech
40
28
552
298.0K
101
Claude Opus 4.5 is actually trained with a TPU
> the price reduction of Opus is only possible because of TPUs. Google has been collaborating extensively with Amazon lately. https://t.co/lqcANLtw2y
> the price reduction of Opus is only possible because of TPUs. Google has been collaborating extensively with Amazon lately. https://t.co/lqcANLtw2y
22
36
592
76.6K
74
π¨ Grok 4.20 is on Design Arena!
Nickname: Granite
I tested it, and you guys are not going to believe this but it's SOTA for frontend! On the level or Gemini 3.0 Pro and Opus 4.5 for frontend. They got the same data as them, it does as good of a job as them, my jaw is on the https://t.co/pYVxaivAsY
Nickname: Granite
I tested it, and you guys are not going to believe this but it's SOTA for frontend! On the level or Gemini 3.0 Pro and Opus 4.5 for frontend. They got the same data as them, it does as good of a job as them, my jaw is on the https://t.co/pYVxaivAsY
50
38
568
708.2K
93
AI News | AI Prompting and Comparison| Breaking AI news before itβs famous
16.2K Followers
12 Contributions
29.2K
Total Members
+ 16
24h Growth
+ 78
7d Growth
Date Members Change
Apr 1, 2026 29.2K +16
Mar 31, 2026 29.1K +3
Mar 30, 2026 29.1K +5
Mar 29, 2026 29.1K +25
Mar 28, 2026 29.1K +18
Mar 27, 2026 29.1K +11
Mar 26, 2026 29.1K +18
Mar 25, 2026 29.1K +7
Mar 24, 2026 29K +10
Mar 23, 2026 29K +5
Mar 22, 2026 29K +8
Mar 21, 2026 29K +14
Mar 20, 2026 29K +7
Mar 19, 2026 29K β
No reviews yet
Be the first to share your experience!
Share Your Experience
Sign in with X to leave a review and help others discover great communities
Login with XLoading...
Discuss the state of AI from industry insiders and anons π΅οΈββοΈ
Community Rules
Make relevant and useful posts π
Post useful, relevant, informative AI insights
Make sure content is High Quality π
Post high quality, original content
Quote Posts must add useful info β οΈ
Do not self-quote. Quote posts must add useful info.
Do not post Spam ποΈ
No spam, promo / ads, off-topic, hashtags
Check earlier posts π¬
Don't post about same thing if others posts first.
Source the insights π
Provide source directly in the main post!
Verify your information π΅ββ
Do not vague post and verify your information before you post!
Don't share private names / methods π€«
Do not share working methods to access info on unreleased products / services in the scope of AI.