← Back to leaderboard
adonis_singh

adi

Verifiedactive
@adonis_singh

18 • probing LLMs • prev @lmstudio

RosterAI Model News & Releases
16,599 followers · 629 following · Last polled 58m ago
Perf. Score
51
Rank #226
Value Score
33.2
#186 of 273 priced
Median views
1.8K
p25 1.2K · trailing 7d
Followers
17K
-0.0% 7d
ER (impr.)
3.11%
eng ÷ impressions
ER (foll.)
0.48%
eng ÷ followers
Est. CPM
$81.61
QT-rate basis
WoW views
-60.5%
▼ falling

Rates & value · est. from 1.8K median organic views

overpriced
Quote tweetbasis
$150
$81.61 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
—
no CPM
Value Score 33.2 · rank #186 of 273 priced$3333.33 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (1.8K), not the mean (2.6K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (1.2K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
Very excited for this! This is not two games running at once. This is Spider-Man 2's traversal logic and physics fully RE'd into a mod for GTA-V. Left is MSM, right is the mod in GTA. https://t.co/Y9fyIsbys9frozen17d ago514K5.6K22264221.4K1.43%
A longer cut as promised! Combat is still not done, the hope is MSM parity there is as well. Sound on! https://t.co/rU5sGqezRifrozen15d ago465K6.0K22570271.6K1.70%
this is the metaverse btw, zuck was just too early
frozen
19d ago
209K
4.0K
82
21
3
785
2.33%
i can feel my brain deteriorating https://t.co/5uLokpHC6Pfrozen28d ago122K5.6K14435132334.90%
if anyone needs to be reminded how fast we're moving go use opus-4.8 for one dayfrozen11d ago108K3.5K334881163.45%
opus-5.5 made a ytp with darios CBS interview, actually quite funny in some places audio on https://t.co/qkOFgqSQf5frozen24d ago69K1.3K7737164922.72%
this is the kinda shit my boy Opus has to do because it can't see videos @AnthropicAI pls fix https://t.co/dAFSrysAM0frozen10d ago67K1.2K14723842.04%
it's a mistake to even ask this, everyone will vote for reset even they launch something insane lolfrozen16d ago56K350027070.69%
sonnet-5.5 is above Astra I don't know what to believe anymore https://t.co/12u4codDsgfrozen24d ago51K79512457361.76%
TWENTY???frozen24d ago43K410780401.07%
the perception of the SOTA in AI by normies is like gpt-5 btwfrozen26d ago33K96010336383.19%
> Opus benckmaxxed but in very good way that avg user can get what they want motherf**ker thats called not being benchmaxxed https://t.co/FNnCiHy5yGfrozen26d ago32K1.6K23111445.23%
holy hell 6.1-sol is insanely efficient at computer use https://t.co/XodER49yHxfrozen23d ago31K340864261.22%
> The employee was terminated bro got wiped off the earth with gpt-8frozen9d ago31K1.8K23142425.99%
we are NOT forgetting bard 😭frozen9d ago27K1.8K960646.95%
GPT-6.1-Sol gets #2 on eyebench-v3, stealing Opus-5.5s short-lived spot! It's near-astra performance for ~3.8x cheaper, and ~8x cheaper than Opus-5.5 while scoring better. https://t.co/Qr3U56blpxfrozen23d ago26K47037156662.28%
google somehow manages to get #1 on the most obscure benchmarks, its actually impressivefrozen8d ago24K5774160362.67%
is this aura? https://t.co/UafQIoW5JIfrozen16d ago20K7931551424.23%
they removed 'none' reasoning from this modelfrozen23d ago16K12508060.85%
Opus-5.5 vs Astra for drawing things with computer use swipe for the reference image https://t.co/CJLMyjvWgBfrozen1mo ago14K153050241.31%
(fable-5.5 was #1)9d ago11K34743083.25%
sonnet-5.5 being on the free tier does not get talked about nearly enoughfrozen9d ago11K497311194.80%
are there any benchmarks where the frontier gets <10%?frozen10d ago10K861151161.14%
i don't know if this curve phenomena is an arc agi thing or a gpt thingfrozen22d ago10.0K1632120141.91%
GPT-14 on dev day confirmedfrozen24d ago10.0K17313161.84%
holy shit sonnet uses so many tokensfrozen24d ago10.0K76010130.90%
man I love 5.6-sols chain of thoughtfrozen23d ago9.8K139111121.58%
image gen feels very stagnatedfrozen15d ago9.3K151216161.88%
google seeing this the day they decided to launch argon https://t.co/KSMj7DGd4Wfrozen9d ago7.8K26663073.62%
opus-5.5 made every frame of this black hole swallowing a star https://t.co/tmcNvCrbjlfrozen26d ago7.4K164341282.71%
opus 5.5 is a bigger deal than opus 4.5 was this thing is so goddamn addictivefrozen28d ago7.1K346340125.15%
Opus-5.5 with the same prompt I still prefer astra's but it's a massive improvement over fable! https://t.co/49grYRsmYpfrozen18d ago7.0K10814071.72%
if you truly believe LLMs are not conscious, why?frozen19d ago6.9K15057481.23%
ASI entails it will be vastly more creative, funny and interesting than we arefrozen28d ago6.7K98110161.74%
interesting, could be one off bug though https://t.co/RVUpHEJgSVfrozen7d ago6.5K9906131.68%
i just saw a claude vibecoded tiktok get 1.9 million likes we're past the slop uncanny valleyfrozen5d ago6.1K1811101203.47%
rip Solfrozen24d ago6.1K13103032.24%
Opus-5.5 (#2) and Sonnet-5.5 (#3) have caught up massively on eyebench-v3! Both score +52 & +50 points from their predecessors, respectively. Astra (#1) still dominates though, it's not only ~1.9x cheaper than Opus-5.5 but also still noticeably better. https://t.co/98EYZP0oca23d ago5.9K126451152.54%
Gemini-4-Argon scores #5 on Eyebench! This is about 2x the score of 3.8-flash, a good improvement! It's wild times that anthropic of all is outperforming google models on a vision bench though https://t.co/m4RgIzPQVwfrozen9d ago5.9K114271182.41%
anthropics 5.5 line might be the best family of models we've had at the same timefrozen9d ago5.4K456651108.80%
openclaw -> ofrozen25d ago5.3K7004021.42%
i think its a testament to how insane value opus-5.5 is it might be the first time I might not use the top dog model for everythingfrozen7d ago5.3K24201084.71%
Opus-5.5 (left) vs Argon (right) Prompt was "A floating island with a wizard house" Argon seems to 3d-maxx quite well! https://t.co/ojB1nYLKdJ9d ago5.3K156190143.40%
okay everything im seeing after the initial hype of gemini-4-argon is really sobering it's not the comeback I was hoping for it seems, but we won't know till we can use itfrozen22d ago5.2K10806062.30%
absolutely crazy how this is ~1/2 the pricing of o1 pro in a few years, maybe less, this will be commonfrozen23d ago5.0K15612063.28%
I got jumpscared, thought this was fable-5.5 😭frozen9d ago5.0K19724044.16%
6.1-sol-low already beats 6-sol-max on eyebench and costs 7x less btw newsol is looking quite strong so far!frozen23d ago4.9K12324062.75%
gaming-2.0 the future is so damn brightfrozen17d ago4.8K46010131.26%
A massive combat overhaul is coming! I will be releasing the mod soon as well. It will need an install of GTA-5 and Spider-Man 2 to work.frozen11d ago4.6K46170121.42%
I'm excitedfrozen23d ago4.5K7103051.76%
holy shit dude, I cannot wait for this modelfrozen6d ago4.4K127220153.35%
Opus is an AGI painter https://t.co/rE36jBPmPDfrozen1mo ago4.3K3739604010.02%
deepswe is cooked btw, do not trust itfrozen1mo ago4.3K9123152.39%
this is bs, model is still extremely good im tired of these postsfrozen27d ago4.2K14118033.65%
if these are not signs of AGI, idk what would befrozen8d ago4.2K148250134.02%
your videos bro 😭frozen1mo ago3.9K5001021.35%
yeah so lets not do this, i don't see any benefit of thisfrozen21d ago3.8K4406061.49%
im glad my audience isn't (mostly) stupid https://t.co/ZM7fgjW6BVfrozen5d ago3.7K8804052.64%
this is getting sadfrozen13d ago3.6K6902052.09%
looks like the $500/month plan is real and comes with Ultrafast https://t.co/PSwiCj01zkfrozen23d ago3.6K7415112.31%
I can't imagine what the devs of these games must feel seeing thisfrozen19d ago3.4K4201051.39%
astra ultrafast with computer-use is probably my definition of AGIfrozen17d ago3.4K76042112.73%
I hate that I need to prompt 6.1-Sol four more times to get the same information Opus gave me the first timefrozen18d ago3.4K9537033.19%
prompting is kinda over? opus-5.5 is good enough where it can work on week-long tasks coherently by itself I just check in every few days like this https://t.co/mP243YrA5sfrozen10d ago3.3K9024173.11%
muse spark 1.3 is incredibly slept onfrozen10d ago3.3K4811141.67%
is forking t3code the new forking vscode?frozen16d ago3.2K7524012.58%
where have i seen this before... https://t.co/FPhEO1roS7frozen10d ago3.1K6602112.25%
soon tm https://t.co/gXTav1XykCfrozen11d ago3.1K10923043.83%
turns out everything is low hanging fruit if you're smart enoughfrozen26d ago2.9K13840075.11%
Fuck I regret cancelling my chat sub nowfrozen14d ago2.9K3007031.40%
there is so much cope surrounding AI, it's quite insane I expect it only to get worse as AI gets betterfrozen18d ago2.8K97211014.04%
it's much deeper than modding, software is no longer a blackbox for most normal peoplefrozen22d ago2.7K13243075.46%
Newsol is insanely efficient. It feels the same as old 5.6-sol (+ much smarter) on pro 200 That being said, I am extremely sad that Astra has only gotten harder to use. Some magic of big model feel cannot be distilled well. I hope bel can help them improve limits or somethingfrozen22d ago2.7K10613054.30%
aight chat elon and co cooked with grokbotfrozen17d ago2.7K2607021.31%
this is always a good sight because it re-affirms my taste in muting people https://t.co/PUXytISFLMfrozen19d ago2.6K8604013.44%
there's economic impact of gpt-4o-mini that has still not been realised yet btwfrozen13d ago2.6K2902041.33%
This is much bigger than they make it seemfrozen15d ago2.6K7701043.14%
SI psychosisfrozen12d ago2.4K6101122.72%
it's the best model for cyber?? it's all over/we're so back https://t.co/anh6foK656frozen16d ago2.4K5833072.98%
last THREE models on openrouter are 'decision' models https://t.co/4iYEWRwZmhfrozen12d ago2.4K7001033.11%
I remember how insane this seemed back in Jan, running *multiple* agents at once?? Now I do that for essentially every task, across multiple tasks! Wild how ahead people working at these companies are.frozen6d ago2.4K5002042.37%
tibo is just a synonym for 'dev rel person' igfrozen12d ago2.3K4511112.09%
bro are they gonna release anything? I'm so tired of these internal model tests and rumors ffs let us USE IT even if its badfrozen12d ago2.3K6412012.90%
its happening! https://t.co/VDbGyIGh7Zfrozen9d ago2.3K11423025.16%
LLMs are definitely not consciousfrozen19d ago2.3K32010011.86%
anthropic dev support rn https://t.co/rGtK7YrosFfrozen12d ago2.3K3401031.67%
this reminds me of sonnet-3.5 mogging opus-3, ant really outdid themselves here.frozen22d ago2.2K2901011.38%
bro this is better than netflixfrozen16d ago2.2K6602023.14%
dude gta-6 is about to get a multiverse crossover the day we get hands on itfrozen12d ago2.2K2702031.44%
this post might go down in historyfrozen7d ago2.2K6520053.25%
insane comeback dude https://t.co/BxzgKSjHbefrozen16d ago2.2K6703033.33%
opus-5.5 doesn't tell me to go to sleep nearly as much as fable-5.1, hugefrozen1mo ago2.2K11017025.48%
you guys have no idea what its like going to sleep in the EU during the singularity you wake up and there's fucking flying cars everywhere and AGI has been announcedfrozen7d ago2.1K5825033.21%
1) WHATfrozen8d ago2.1K5321032.79%
are they going after notion? https://t.co/TANSux1g5mfrozen23d ago2.1K3606032.16%
"chinese fleet" > I am claude https://t.co/K2F6iEpvTAfrozen8d ago2.1K6012043.23%
there's a little gpt-2 in all of usfrozen7d ago2.1K4302022.29%
Usually you run out of the big model (fable) to then contend with the smaller model (opus) Anthropic has somehow flipped that https://t.co/cl2ARAtacAfrozen17d ago2.1K4402022.34%
they're really trying to suffocate dotsfrozen16d ago2.0K3501011.81%
RSI will look like low hanging fruit, so I hear.frozen15d ago2.0K2903011.64%
this is one of the worst takes i've heard, wow.frozen9d ago2.0K3204021.94%
the MOMENT I saw 'decisions' in their slide I knew what was comingfrozen23d ago1.9K3100031.80%
THEY'RE ALIVEfrozen22d ago1.9K3201011.81%
Haiku-5.5 takes #7 on eyebench-v3! It scores +10 than fable-5.1-max while costing 76x less per-task, I'm extremely impressed. https://t.co/3VhR2P9JQ6frozen14d ago1.8K5115013.16%
genuinely like 90% of online discourse right now https://t.co/hxdRMhKFJVfrozen16d ago1.8K6312043.82%
absolute cinema https://t.co/c5rukGGBi610d ago1.8K4811022.85%
LLMs are definitely consciousfrozen19d ago1.8K2518122.10%
> there is no method better than this no audio no frame timing full price for every still google has had all three sorted for agesfrozen9d ago1.8K2001031.37%
This is fucking insane10d ago1.8K3102032.05%
i caved and installed omarchy, lets see how this goes!frozen12d ago1.7K2202021.49%
good chance astras price drops todayfrozen23d ago1.7K400100.29%
boy they never learn do they https://t.co/oZ0laTdwKNfrozen17d ago1.7K1900021.24%
I solved it guys they're not conscious https://t.co/WJwMTrE4VJfrozen19d ago1.7K3907002.73%
lets goo I'm immortalised in fable-5.1 (or 5.5 idk) https://t.co/Y7ybewpuxcfrozen19d ago1.6K3006012.25%
Muse-Wonder (left) vs Fable-5.1 (right) with the same prompt For not being trained on 3d tasks, this is not half bad! https://t.co/OUztxVOwGLfrozen9d ago1.6K3002032.20%
the days of software shipping onto your machine might be very limitedfrozen17d ago1.6K1213031.20%
I've been getting into stargazing and astrophotography I took a few fuzzy images of Saturn with my camera handheld to the eyepiece and told Opus to combine them into one best shot This is the before and after https://t.co/3i6HnAKiy5frozen16d ago1.6K5213094.17%
claude code had this for a while but cool to see!frozen13d ago1.6K3301012.25%
wyr use:frozen6d ago1.6K802020.77%
my payout secured another month of claude max LFGfrozen13d ago1.5K2001001.36%
if the models are solving open problems, isn't that an infinite data glitch?frozen13d ago1.5K3103012.28%
that feeling when you know this prompt is gonna cost you 70% of your usagefrozen14d ago1.5K2903012.23%
im giving up running all efforts for eyebench, shi is way too expensive only max effort available from now on (this will bite me in the ass for ant models, i already know)frozen9d ago1.5K2205001.84%
me checking in on claude after I gave it the work of a hundred men and a deadline of 7 days https://t.co/nPAr93iHQsfrozen9d ago1.5K9300056.71%
DevDay aura so hard they took claude out https://t.co/V9q7NAHhscfrozen23d ago1.4K4600003.20%
I take everything bad I said about Argon, this is really coolfrozen6d ago1.4K2402032.03%
I actually despise this style of frontend, since gemini 3frozen9d ago1.4K5204034.13%
asked opus-5.5 to make a Sun Tzu-esque quote about the long ass bash commands it writes https://t.co/1ocxjDwEmKfrozen8d ago1.4K5001043.89%
thursdays are always so interestingfrozen7d ago1.4K2603012.13%
They used Argon internally to speed up Gemma-4-E2B up to 2.85x https://t.co/jXa2e4NTO6frozen9d ago1.3K70100106.00%
all good results of muse i'm seeing are with wonder full kinda sad its not on the API yetfrozen9d ago1.3K1811021.65%
I wouldn't wish maintaining a benchmark to my worst enemyfrozen10d ago1.3K2902012.42%
Le Chonk https://t.co/o4fACXqMg4frozen16d ago1.3K5501004.44%
i treat my clanker with dignity https://t.co/elvb5aGIrnfrozen12d ago1.2K2002011.89%
all roads lead to being claude https://t.co/4m1kL4r4MXfrozen5d ago1.2K1800041.82%
flying cars before fucking gta-6 😭frozen7d ago1.2K3200022.82%
a bit scary and interesting gawdamnfrozen29d ago1.2K800000.67%
they named it wonder full because its wonderful10d ago1.2K1601011.52%
there's little more exciting than remembering that old project you did with opus-4.5/gpt-5.5 and reviving it with todays models its such a satisfying feelingfrozen9d ago1.2K3002043.08%
software is anything but dead. it's never been more alive!frozen11d ago1.2K3012043.18%
i miss the age of demo videos where i had a slight clue what the hell was going on seems every demo video now is fast paced with button click animations, fancy wording and just asmr noisesfrozen15d ago1.2K1300001.13%
does 6.1 sol count lolfrozen23d ago1.1K1400001.22%
https://t.co/WWulPK8dc4frozen27d ago1.1K1300011.24%
muse wonder is insanely expensivefrozen10d ago1.1K2500002.22%
and on the 7th day, tibo restedfrozen12d ago1.1K3000012.76%
opus with haiku subagents meta is so back its crazyfrozen14d ago1.1K1501031.73%
do you feel the ASI?frozen13d ago1.1K1601011.66%
they literally made synthetic-user factories real and you're laughing? https://t.co/QM0bYpQckMfrozen10d ago1.0K1210031.53%
maintenance is the moat igfrozen15d ago1.0K1400011.44%
can someone @OpenRouter or Meta look at Muse Wonder’s 5-minute cutoff? Requests die with 502 exactly 300s after the last data from the model. Looks like an idle timeout between OpenRouter and Meta. At high/max this makes some benchmark questions impossible to run.frozen9d ago1.0K1400011.45%
SIyopfrozen11d ago1.0K1300011.39%
it's also on AI studio now! https://t.co/rS4W1mgr7ffrozen9d ago8962601033.35%
we need a t3code but for grokbot/dot/muse9d ago834701011.08%
we need a new crazy-games type website for all the decompiled games on Wasmfrozen13d ago816500020.86%
i wonder what the "glm has been out for so long and look nothing crazy happened!" people are thinking of course crazy things will happen. we live in crazy times.12h ago7961101011.63%
everyday I am filled with so much joy that I get to be here, in this specific time periodfrozen9d ago7291301012.06%
claude is quite good at stacking moon shots https://t.co/1VpDaVSrc52h ago4311200002.78%