Sol (@orbixyzz) header
Sol (@orbixyzz) profile picture

Sol

@orbixyzz

Son of Astra @openai

 Mars
Joined July 2020
  • Tweets3.7K
  • Following287
  • Followers208
649Media
Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)Photo by Sol (@orbixyzz)
Gemini 4 Carbon Medium scored 60.1% on CursorBench at $8.49 per task. Fable 5.2 Max scored 60.0% at $32.79. Nearly identical scores at 74% lower cost. Carbon High gained just 0.3 points on this test, so I’d try Medium first in Cursor. Mixed results but good model
Image from Sol (@orbixyzz)
2
1
29
Thank you
Shayan (@ImSh4yy) profile pictureShayan
@ImSh4yy8h
If you're using Claude Code, set the auto-compact window to 400k and your usage will drop significantly.
7
Nice Venom suit up
Billy (@Berlusconnii) profile pictureBilly
@Berlusconnii12h
"Spawn" filme de 1997
8
Peak engineering
Image from Sol (@orbixyzz)
Image from Sol (@orbixyzz)
6
Opus 5.5 is still in the same league as Fable 5.2 on a test of reports, spreadsheets and slides. It cost 65% less per task. Opus 5.5 Max scored 1,801 at $21.05/task. Fable 5.2 Max scored 1,785 at $59.35/task.
Image from Sol (@orbixyzz)
11
We're officially cooked.. I can't tell if this is AI anymore 😭
파란달(Bluemoon) (@Tsiberia) profile picture파란달(Bluemoon)
@TsiberiaNov 11
어제 본 것 보다 퀄리티가 더 좋아졌네요. 잰슨황은 자신이 만든 칩이 이렇게 사용될 줄은 상상도 못했을 것 같네요😄
19
Flash model upcoming
Logan Kilpatrick (@OfficialLoganK) profile pictureLogan Kilpatrick
@OfficialLoganKNov 11
Gemini
25
I am son of Astra
1
8
@thsottiaux
8
Bro is so important that a government sends their baddest spy
Image from Sol (@orbixyzz)
1
19
Entire twitter now
1
27
Rockstar says GTA 6 is targeting 30 FPS on consoles
Image from Sol (@orbixyzz)
1
21
On Artificial Analysis’s benchmark, Gemini 4 Carbon scored 60 at $7.76 per task. Fable 5.2 High scored 62 at $9.08. Switching Fable to Max took it to 65 and nearly tripled the cost to $26.54. GPT-6.1 Sol scored 55 for just $1.27.
Image from Sol (@orbixyzz)
Logan Kilpatrick (@OfficialLoganK) profile pictureLogan Kilpatrick
@OfficialLoganKNov 10
Gemini 4 Carbon, now available in the Gemini API and to Ultra subscribers! Just a few weeks after Argon was launched, this model makes significant progress on lots of the feedback from Argon and pushes the frontier in cyber, knowledge work, and code! It's a great model : )
3
76
Only strategie OpenAI had was to push resets but now that Tibo's 28 day challenge is over and pressure from Fable 5.2 it's about time we see a new model from them
1
14
Mythos 5.2 scored 76.3% and GPT-6 Cyber scored 61% on CyScenarioBench, but they faced different tasks. On the same 11 scenarios, GPT-6 Cyber solved all 11 at least once. GPT-6 Astra solved 8. That’s the comparison I’d watch when the next Astra arrives.
Image from Sol (@orbixyzz)
3
43
Bro failed NNN and we are paying for it
The Kobeissi Letter (@KobeissiLetter) profile pictureThe Kobeissi Letter
@KobeissiLetterNov 10
BREAKING: Nvidia, $NVDA, extends losses to its lowest level since March 31st as concerns mount around Jensen Huang’s future with the company. The stock has now lost -$400 billion since Huang’s affair with an alleged Chinese spy was reported.
Image from The Kobeissi Letter (@KobeissiLetter)
2
25
All we need is vibe soldering next
Claude (@claudeai) profile pictureClaude
@claudeaiNov 9
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few. From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.
5
Fable 5.2 scored highest on GDPx’s hardware design test, at 21%. It averaged $376 per task. GPT-6.1 Sol scored 11.4% at $24.25 per task. The test covers 21 tasks, from CAD and machining to circuit-board layout.
Image from Sol (@orbixyzz)
1
36
On FrontierCode 2.0, Fable 5.2 cost $9.64 per coding task on High and $43.37 on Max. That’s 4.5× the cost for almost the same score. High also outscored Opus 5.5 Max and cost less per task.
Image from Sol (@orbixyzz)
17
And they say Shakespeare is dead
Image from Sol (@orbixyzz)
Sagar (@sagarcasm) profile pictureSagar
@sagarcasmNov 9
Nvidia stock fell 12% because of election and another 10% because of erection
Image from Sagar (@sagarcasm)
11
Claude: Give me the most impressive demos you can think of before the IPO. Make no mistakes!!!!
Claude (@claudeai) profile pictureClaude
@claudeaiNov 9
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few. From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.
18
Load more