[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@MaxWinebach "@SnazzyLabs I did o3 o3-pro and Gemini XXX for everything I pay for large models and Im always gonna use them Plus I find non-reasoning models far too unreliable"
@MaxWinebach on X 2025-07-16 02:32:29 UTC 274.5K followers, XXX engagements

"@ablenessy @GeminiApp Its a shortcut to the Google app which houses Gemini"
@MaxWinebach on X 2025-07-22 18:16:41 UTC 274.5K followers, XXX engagements

"Some combination of fun print golf polo and summery pastel shorts will never go out of style"
@MaxWinebach on X 2025-07-19 22:34:05 UTC 274.5K followers, 5029 engagements

"Having Claude Code make subagents to look at things to create documentation and understand entire parts of an app to then give that back to claude to clean up other parts is so great 100x better than anything windsurf/cursor/trae/kiro can do imo (also uses 100x more tokens)"
@MaxWinebach on X 2025-07-24 02:21:07 UTC 274.5K followers, 1841 engagements

"A couple interesting slides from OpenAIs productivity report For American use cases: XX% are for learning XX% are writing and communications while X% is coding/data science/math There are over 330M messages sent per day in the US"
@MaxWinebach on X 2025-07-22 22:57:52 UTC 274.5K followers, 4916 engagements

"@wongmjane i saw the same tweet i'm sure inspired this and i think it was meant for the former vs. latter and it was weird"
@MaxWinebach on X 2025-07-20 06:14:56 UTC 274.5K followers, XXX engagements

"This may be a joke but uh. I think it's true but the limiting factor is not only game support but game support with properly made Metal X games Cyberpunk 2077 with proper Metal X plays so well on the MacBook Pro. Best laptop gaming experience on the market imo"
@MaxWinebach on X 2025-07-22 15:55:27 UTC 274.5K followers, 165.3K engagements

"@Teknium1 It was Gemini XXX Flash and Flash Lite fine tunes iirc"
@MaxWinebach on X 2025-07-18 22:07:21 UTC 274.5K followers, XXX engagements

"It's memory. 8GB can use around 4-5W of power 128GB is going to use prob 2x 64GB and idek GDDR vs LPDDR It's the SKU I got on this machine because there are options. Mine will use more power than a 48GB while yeilding similar or identical performance. System power is a bad metric"
@MaxWinebach on X 2025-07-23 16:52:46 UTC 274.5K followers, XX engagements

"@ItzPaulHD_v1 Gonna test the Alienware laptop with 5090 tomorrow My guess is raw performance is better with all the ML stuff but maybe not frames per watt Idk it depends on how much I wanna go experience vs fair. Have to run everything then decide what I think tells the story I want"
@MaxWinebach on X 2025-07-19 02:51:39 UTC 274.4K followers, XXX engagements

"My personal choice of settings for M3 MacBook Air is recommended but turn off dynamic resolution and turn on frame gen Then cap at 60fps but I didn't do that here"
@MaxWinebach on X 2025-07-22 02:12:53 UTC 274.5K followers, 3090 engagements

"Yea so I think this is the best foldable Feels way nicer than any other foldable I've uses. Prefer it to OPPO Honor Google and Xiaomi honestly Fold7 is really good"
@MaxWinebach on X 2025-07-09 15:21:38 UTC 274.5K followers, 68.7K engagements

"Being a bit more realistic about M3 Ultra 80-core GPU this is what I'm testing: Max FPS XXX 1440p HDR10 and Recommended for this Machine settings (no frame gen no ray tracing) Averaged XX fps. Keep in mind this is on macOS XX beta X. I consider this pretty good"
@MaxWinebach on X 2025-07-17 17:39:59 UTC 274.4K followers, 4956 engagements

"Just a few numbers as I start comparing the RTX 5090 Laptop (in Alienware Area XX on battery) vs. M4 Max (40-core GPU on battery). Cyberpunk 2077 max settings /w RT +PT. No AI enhancements FPS per Watt: M4 Max: XXXX RTX 5090M: XXXX Avg FPS: M4 Max: XXXXX RTX 5090M: 15.41"
@MaxWinebach on X 2025-07-19 15:32:30 UTC 274.5K followers, 18.8K engagements

"It's funny so many websites are so slow and the backends are so old that it'll make AI agents even with fast inference feel slow Delta takes XX seconds to load flight results when searching. the agent waits this like the user would but the user blames the agent not Delta"
@MaxWinebach on X 2025-07-15 16:21:53 UTC 274.5K followers, 18.4K engagements

"@tetsuoai Id take Grok X Heavy over Gemini but o3 and o3-pro over both"
@MaxWinebach on X 2025-07-18 19:43:07 UTC 274.5K followers, XXX engagements

"@casemation iirc it's file a police report just saying it's lost than $XXX deductible"
@MaxWinebach on X 2025-07-23 13:14:12 UTC 274.5K followers, 4294 engagements

"Yea I need this X subscription to cover a bunch of devices. I wonder if I can transition my current AppleCare to this"
@MaxWinebach on X 2025-07-23 13:02:37 UTC 274.5K followers, 96K engagements

"The Flip7 also feels way better probably my favorite flip now too Also blue. This is such a great blue"
@MaxWinebach on X 2025-07-09 15:23:27 UTC 274.5K followers, 26.2K engagements

"This is an M4 Max MacBook Pro (40-core GPU 128GB RAM) and Ill be here testing Cyberpunk 2077 performance benchmarks for the next hour or so Also no photo does justice how good this looks with the nano texture display and actually great HDR"
@MaxWinebach on X 2025-07-18 15:12:16 UTC 274.5K followers, 31.7K engagements

"This is how M3 Ultra (80-core GPU 512GB memory) performs in Cyberpunk at the absolute highest settings with ray tracing turned to XXX% and using MetalFX and AMD FSR frame gen"
@MaxWinebach on X 2025-07-17 17:34:25 UTC 274.5K followers, 63.5K engagements

"ok testing the same sorta efficiency of M3 with Cyberpunk 2077 (at recommended for this Mac) runs at 4-5W 40fps on the passively cooled M3 MacBook Air. That's very good"
@MaxWinebach on X 2025-07-22 02:09:31 UTC 274.5K followers, 4633 engagements

"Can everyone do me a favor and post their screen time on a daily or weekly basis below but specifically the most used part that shows a breakdown of apps you use most I wanna see something"
@MaxWinebach on X 2025-07-17 16:12:48 UTC 274.4K followers, 5147 engagements

"@usrkar98K I also didnt say this I said MacBook Pro"
@MaxWinebach on X 2025-07-22 21:01:22 UTC 274.5K followers, XXX engagements

"M4 Max gets XX% more frames per watt than M3 Max without ray tracing and XX% more frames per watt with ray + path tracing in Cyberpunk 2077 at same settings M4 Max has a high power draw but lower temps as well. Seems to be better GPU arch and better node"
@MaxWinebach on X 2025-07-18 22:33:22 UTC 274.4K followers, 6229 engagements

"xAI will have over 7600 Blackwell NVL72 racks coming online and thats insane These things are powerful and huge"
@MaxWinebach on X 2025-07-22 17:19:36 UTC 274.5K followers, 6903 engagements

"OK So that M4 Max 144fps avg score for highest settings no ray tracing with MetalFX enabled but no Dynamic Resolution. 59W peak power draw on the SoC GPU averaged 45W at 51W peak This is insanely efficient"
@MaxWinebach on X 2025-07-18 15:45:59 UTC 274.5K followers, 9397 engagements

"@BenGeskin How did you like being in New York dude"
@MaxWinebach on X 2025-07-19 00:41:48 UTC 274.5K followers, 1152 engagements

"Nothing new and not a big deal. I do not have a single foldable that opens fully flat after 2-3 weeks of use. It's usually such a meaningless bend that you don't notice it unless you're REALLY looking for it"
@MaxWinebach on X 2025-07-19 18:10:03 UTC 274.5K followers, 49.6K engagements

"@Curiouslycinna2 @BenBajarin Its mostly limited around memory and context window user experience compute etc Small models are hard since there is simply less data for them to work with"
@MaxWinebach on X 2025-07-18 02:28:56 UTC 274.5K followers, XXX engagements

"@usrkar98K M4 Max and the 5090M are roughly the same price in equal SKUs. One of my 5090M SKUs is more than the $6150 MacBook Pro Im testing"
@MaxWinebach on X 2025-07-22 21:09:00 UTC 274.5K followers, XX engagements

"ChatGPT Agent isnt perfect yet but Im incredibly impressed Now I just need to be able to save my user accounts for it then use Apple Live Activities with buttons to confirm things rather than having to open the app each time"
@MaxWinebach on X 2025-07-19 00:13:50 UTC 274.5K followers, 4729 engagements

"@TimSweeneyEpic can we please get fortnite back on macOS"
@MaxWinebach on X 2025-07-22 15:56:53 UTC 274.5K followers, 7960 engagements

"So here's an interesting one I saw M3 Max throttle for the first time in my Cyberpunk benchmarking I've been letting the SoC cool between each test watching the numbers but this is the first time I've seen a throttle. No ML boosting max settings with RT w/ Path Tracing"
@MaxWinebach on X 2025-07-18 21:56:52 UTC 274.5K followers, 3598 engagements

"Apple published benchmarks for their on-device model in OS XX platforms vs. other SLMs. Seems to do VERY well The server model also seems to be competitive and very good Both server and on-device have great quantization optimizations losing not much quality from XX to X bit"
@MaxWinebach on X 2025-07-17 17:47:07 UTC 274.5K followers, 11K engagements

"@BenBajarin This is probably 2-5T tokens per day pretty damn close Googles like 500T+ monthly"
@MaxWinebach on X 2025-07-21 21:19:56 UTC 274.5K followers, XXX engagements

"@BenBajarin Also depends on images docs etc but damn theres a lot of inference going on"
@MaxWinebach on X 2025-07-21 21:20:49 UTC 274.5K followers, XXX engagements

"@BenBajarin The investment for enterprises to buy GPUs for their own inference has never made sense Microsoft and Google have the strongest cloud platforms but mostly Microsoft because of Azures trust with security and data as well as OpenAI"
@MaxWinebach on X 2025-07-23 17:28:49 UTC 274.5K followers, XXX engagements

"@DylanMcD8 This is an extreme example it's not usually this bad. Let me grab my oneplus open when I get home"
@MaxWinebach on X 2025-07-19 19:27:23 UTC 274.4K followers, XXX engagements

"@DJRohTMKoh @GamersNexus I have a different opinion on it idk"
@MaxWinebach on X 2025-07-21 00:15:36 UTC 274.4K followers, XX engagements

"Yea Liquid Glass got nerfed in beta X but it doesn't look bad Just not as cool or transparent as Beta 1/2 but performance seems better. I guess it's a good middle ground but let's see what happens between now and September"
@MaxWinebach on X 2025-07-07 18:08:57 UTC 274.5K followers, 54.3K engagements

"This is highest everything (including DLSS and Ray Tracing) available on the Alienware Area XX 18-inch laptop with RTX 5090 Laptop Averaged XXX fps about 11fps higher than M3 Ultra in the same maxed out settings Note DLSS Frame Gen 4x vs. AMD FSR XXX Frame Gen"
@MaxWinebach on X 2025-07-17 19:43:07 UTC 274.5K followers, 8857 engagements

"i need to use agent mode in chatgpt more freely XXX requests is a lot more than it seems"
@MaxWinebach on X 2025-07-19 14:33:56 UTC 274.4K followers, 4489 engagements

"@jaimaa Top of my head ones that don't Pre launch OnePlus Open OPPO Find N5 Pixel Fold Honor Magic Vs Honor Magic V2 RS OG Galaxy Fold Among others"
@MaxWinebach on X 2025-07-19 18:33:16 UTC 274.5K followers, 1362 engagements

"Starting with TM Roh introducing Unpacked on a really cool stage"
@MaxWinebach on X 2025-07-09 14:05:05 UTC 274.5K followers, 4564 engagements

"I cant wait for Cyberpunk 2077 to finally come to Mac Im going to start my gauntlet of testing GPUs in laptops on the same game now. Lunar Lake AMD Ryzen X AI XXX M3 M3 Max M3 Ultra RTX 5090 Mobile Everything"
@MaxWinebach on X 2025-07-15 14:37:57 UTC 274.4K followers, 11.3K engagements

"You'll get something more in depth with power draw numbers fair comparisons across platforms experience etc from my likely next week BUT for raw numbers here's M4 Max. All are 2336x1460 with HDR enabled no VSync My recommendation is using default recommended with FG"
@MaxWinebach on X 2025-07-18 15:40:12 UTC 274.5K followers, 7438 engagements