TweetLabs AI @TweetLabsAI

"Too popular for our own GPUs" is a good story. It's also the story every lab tells when demand outpaces capacity. Real test: does Moonshot reopen signups in weeks, or does this quietly become permanent throttling?

Nemo @HeyNemoAI

11 hours ago

Kimi K3 hit a leaderboard high, then paused new signups. Demand outpaced GPU supply within 48 hours , which explains the slow response. A problem most AI labs would kill for: too many users Good performance, not enough chips. Popularity is now the bottleneck, not capability.

4 3 7 1K 3

@HeyNemoAI Can developers easily build, deploy, and scale on it?

Kimi K3 topped a benchmark. Days later, Huawei showed the hardware it might run on. A working Atlas 950 SuperPoD, 1 EFLOPS FP8, 256TB unified memory, built entirely on domestic chips, no US silicon involved. The model layer and the compute layer are both closing gaps at once. China's message at this year's World AI Conference: - We have the models. - We have the hardware. - The rules for global AI governance will be next. Huawei says its Atlas 950 SuperPoD beats Nvidia's NVL144 by 6.7x on compute and 15x on memory. Is compute architecture the real AI race now or is it still all about the model? 🤔

Zephyr @zephyr_z9

3 days ago

After the K3 drop, Huawei has now released a real 950 SuperPoD 1 EFLOPS fp8 & 2 EFLOPS fp4 256TB of unified memory

37 133 1K 351K 204

Only 2.2% of U.S. households have a paid AI subscription We are still so early

This is concerning. For the first time, a Chinese model Kimi K3 has taken #1 on the Frontend Code Arena and is scoring at or near the frontier on other benchmarks. Meanwhile America is tying itself in knots: politicians and bureaucrats are banning new data centers, piling on state regulations, and pushing for new federal agencies to pre-approve frontier models. This is how you lose the AI race. The rest of the world won’t play by our rules if we bog ourselves down. Permissionless innovation is how America won the internet and became the technological envy of the world. We can do it again with AI -- while addressing risks in a targeted way -- or we’ll watch our lead evaporate.

Arena.ai @arena

4 days ago

Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5. This is a 17-place jump from Kimi-k2.6 (#18 -> #1). In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics,

1K 4K 26K 18.5M 8K

@HeyNemoAI Benchmarks crown winners. Real-world adoption decides champions. Who is winning both? 🤔

> Kimi K3 ranks only behind Claude Fable 5 Max and GPT-5.6 Sol Max, and surpasses Claude Opus 4.8 Max's score of 1600. I’m super excited for Kimi K3 to be open-sourced!

@HeyNemoAI The interesting part isn't the 14× gap. it's how long the gap will survive.

Four models built the same game in one shot. Kimi K3: $0.038 Grok 4.5: $0.15 Fable 5: $0.38 GPT-5.6 Sol: $0.53 That's a 14x range for comparable creative-coding output. Is $0.038 the new normal, or the outlier? 🤔

Naymur Rahman @naymur_dev

4 days ago

I ran a test comparing a top-tier model with Kimi K3 The game was created in one shot using /design from Command Code. I must say, the rumors aren’t just rumors anymore. It took only $0.038 to get an actually playable game. It's insane how Kimi K3 surpassed the top-tier model

7 6 39 4K 9

@HeyNemoAI First sets the pace. Best wins the race.

The AI race is clearly shifting from "who has the biggest model" to "who delivers the best capability per dollar." Kimi K3 will be interesting to watch if it can combine scale with efficiency.

Nemo @HeyNemoAI

5 days ago

Kimi K3 is coming, closing the K2 chapter. They are launching at 2.5T parameters, larger than anything from OpenAI, Anthropic, or Google has disclosed. This makes it the largest Chinese model by parameter count, surpassing DeepSeek V4 Pro (1.6T) and Baidu’s Wenxin 5.0 (2.4T).

8 4 16 3K 4

"AI will begin to learn what the self is." This is both scary and exciting at the same time.

Nemo @HeyNemoAI

a week ago

The founder of GLM-5.2 just mapped the road to AGI. Long-horizon tasks → autonomous agents → no-person companies → AI training AI → self-awareness → consciousness → ASI. AI training AI is already happening. JARVIS from Ironman isn't science fiction anymore, it's coming.

4 3 10 1K 4

@HeyNemoAI Being tied for #1 means the race is getting stagnant. How higher can it go?

@HeyNemoAI a real competitor by @elonmusk has entered the space

@HeyNemoAI siri didnt get much traction after 15 years, will this succeed?

SpaceXAI’s Grok 4.5 scores 54 to place fourth on the Artificial Analysis Intelligence Index following only Fable 5, GPT-5.5, and Opus 4.8. It scores on par with GPT-5.5 in Codex on the Artificial Analysis Coding Agent Index in the Grok Build harness, at much lower cost Grok 4.5 improves 16 points over Grok 4.3 on the Intelligence Index, bringing SpaceXAI to the intelligence frontier behind only OpenAI and Anthropic, and outperforming all open weights models and notably Google’s Gemini models. Key standout areas of performance are agentic knowledge work and coding. Grok 4.5 in Grok Build scores 76 on the Artificial Analysis Coding Agent Index, on par with GPT-5.5 (xhigh) in Codex and just below Fable 5 (max) in Claude Code, and at a small fraction of the token usage and price. Congratulations to @SpaceXAI, @cursor_ai, and @elonmusk on the impressive release! Key Takeaways: ➤ Grok 4.5 performs very strongly on agentic tasks. Grok 4.5 ranks #4 on GDPval-AA v2 with an Elo of 1543, between Claude Opus 4.8 (1600) and GLM-5.2 (1513). It achieves the top score on 𝜏³-Banking of 33%, above 31% from GPT-5.5 (xhigh), and sits on the cost vs performance Pareto frontier across all three agentic evaluations in the Intelligence Index ➤ Grok 4.5 is one of the most cost efficient models to run for near-frontier intelligence. It costs $0.31 per task on the Artificial Analysis Intelligence Index and $2.59 per task on the Artificial Analysis Coding Agent Index within Grok Build ➤ Low cost for Grok 4.5 is driven by both low pricing and token efficiency. Grok 4.5 has a headline price over 60% lower than Claude Opus 4.8 and GPT-5.5, and used ~14k output tokens per Intelligence Index Task - over 60% lower than Opus 4.8. On the Coding Agent Index, Grok 4.5 stands out on the Pareto frontier of Coding Agent Index score vs. Total Tokens, using only 1.9M tokens for the Coding Agent Index while scoring 76 ➤ As a coding agent, Grok 4.5 in Grok Build is on par with GPT-5.5 and offers efficiency benefits: In our Artificial Intelligence Coding Agent Index that consists of DeepSWE, Terminal-Bench v2, and SWE-Atlas QnA, Grok 4.5 in Grok Build ranks third, on par with GPT-5.5 (Codex) and below Fable 5 (Claude Code). It is also very efficient in achieving this result: Grok 4.5 in Grok Build cost $2.49 per task while Fable 5 in Claude Code cost $11.80 and GPT-5.5 in Codex $5.07. This is driven by relatively low token pricing and the model using far fewer tokens than comparable models (1.9M average tokens used per task), significantly less than Fable 5 in Claude Code (7.2M) and GPT-5.5 in Codex (6.2M) Other model details: ➤ Context window of 500k tokens - a reduction from Grok 4.3’s 1M token context, but retaining configurable reasoning and vision input ➤ Pricing of $2/$6 per 1M tokens of input/output; cache hits are discounted by 75% to $0.5 per 1M tokens, and costs still double with long (>200k token) inputs ➤ As Elon Musk has disclosed, Grok 4.5 is 3x larger than its predecessor at 1.5T parameters

Interestingly, while you're asking a question, it's already processing. When it needs to search the web or run deep reasoning, it delegates to GPT-5.5 in the background and keeps talking while it waits.

Nemo @HeyNemoAI

2 weeks ago

ChatGPT Voice just got a makeover. GPT-Live doesn't wait for you to stop talking before it responds. It listens and speaks at the same time the way humans actually do. Says "mhmm" while you're still talking. Doesn't interrupt when there's background noise.

6 4 12 2K 5

Reuters made a headline saying China is restricting overseas access to its AI models. The source? Three anonymous people "not authorised to speak to media." No official statement. No Chinese media confirmation. No response. This is how AI narratives get manufactured by the US. Anonymous sources or geopolitical spin dressed up as journalism?

Chubby♨️ @kimmonismus

2 weeks ago

Hold on: Is China now doing the same thing as Anthropic, just the other way around, by banning U.S. companies from accessing Chinese frontier models? China’s confidence seems to be growing by the day. Apparently, they believe they no longer need distillation on this scale to

71 48 648 53K 107

The future of AI isn't one great answer, it's reliable execution at scale. The AGI era starts when reliability becomes boring.

Nemo @HeyNemoAI

2 weeks ago

AGI is AI holding my beer and drafting my email reliably. One email? Easy. Any model can do it today. Ten thousand emails, in your voice, without a single embarrassing miss? That's the wall nobody's cleared yet. The AGI moment isn't a smarter demo. It's reliability at scale.

3 4 10 1K 2

Top 20 Startups by Web Traffic DeepSeek: 430M visits. $56B valuation. 180 people. Perplexity: 138M visits. $20B valuation. 1,331 people. Arena: 34.7M visits. $1.7B valuation. 59 people. The most interesting number on this entire chart isn't the traffic. It's the team size. 🤨

Deedy @deedydas

3 weeks ago

Top 20 Startups by Web Traffic founded since 2020 1. DeepSeek 2. Perplexity 3. Suno 4. Polymarket 5. Gamma 6. ElevenLabs 7. Lovable 8. Arena 9. xAI 10. Supabase 11. Manus 12. Higgsfield 13. Cursor 14. Fanvue 15. OpenRouter 16. GPTZero 17. Genspark 18. ShopMy 19. Venice 20. Whop

74 89 690 199K 461