@viralasylum2@LiuInTheShadows CXMT is targeting 'legacy' DRAM, not AI HBM. The vast majority of the market by volume is still standard memory. As long as demand for HBM is high, the oligopoly will be mostly fine... until they are not.
China is also a huge market for them... Factory of the world, remember?
>Antigravity also really needs to be a desktop app.
What does this mean? Antigravity is a DESKTOP app.
This is the second time I have seen you say the same thing. Do you really know what Antigravity is or are your posts just AI slop?
Google can't just catch up any more. They need to leapfrog.
OpenAI has GPT-5.6 in the pipeline. Claude's next model is coming. By the time Google matches today's benchmark, everyone else will have moved on. For me, Gemini is a clear third. Sure, Gemini has audio and video input that others don't, but the text model is lacking. Antigravity also really needs to be a desktop app. I rarely choose Gemini over OpenAI and Anthropic models. If ChatGPT adds omni support, I'll have no need for Gemini at all.
I/O is their shot to prove they're competing to win. Right now I'm not convinced most people wake up excited to use Gemini over the alternatives. Next week needs to change that.
@BitsByNick@scaling01 > Per-token pricing has doubled from GPT-5.4 to $5/$30 per 1M input/output tokens. However, a ~40% token use reduction largely absorbs the hike - resulting in a net ~+20% cost to run our Intelligence Index.
At least read before trying to post a dunk. Makes you look dumb.
@olegpustovit@yacineMTB Codex Team: We have made 5.3-Codex 10% faster and that is on top of the earlier 60% improvement.
Whole TL: Wow. Codex has gotten insanely fast. They have basically caught up to Claude.
This guy:
The biggest AI model showdown is here. 🚀
Kling 3.0 vs Seedance 2.0 vs Sora 2 vs VEO 3.1
One prompt. Four models.
Who delivers the best results?
Cast your vote 👇
@dipeshdev_ @shortpermansu It wouldn't make sense for them to tarnish their reputation over something like that.
You could've just waited a bit for the result instead of wrongly declaring Opus #1 even after 5.3-Codex was already out... or just add a disclaimer 🤷♂️
Opus 4.6 result is not verified, BTW.