Grok Imagine
Grok ImagineDéborah: We are beings of light, and Grok reveals this light.✨
Model archive
AI models 28 tracked
Every primary-source story across every tracked model. Filter by clicking a chip.
Grok Build upgrades
Grok Build upgradesX Freeze: Grok Build just received another major update, introducing powerful conversation navigation, enhanced enterprise controls, smarter terminal workflows, and extensive reliability improvementsRelease Notes:…
> the $650B in capex that you mention is mostly yet to come online Right now, Anthropic has 50x-150x more compute than Moonshot they better hurry up w…
> the $650B in capex that you mention is mostly yet to come onlineRight now, Anthropic has 50x-150x more compute than Moonshotthey better hurry up with…
ChatGPT Work is great for tasks in your personal life:
ChatGPT Work is great for tasks in your personal life:Andrew Ambrosino: the “work” in ChatGPT Work is describing what the agent is doing for you– you…
RT tokenbender: in a parallel world, karpathy has posted about how amazing kimi k3 is and how he is excited about using it to build banger writer/stor…
RT tokenbenderin a parallel world, karpathy has posted about how amazing kimi k3 is and how he is excited about using it to build banger writer/storyteller…
Zhipu grew their ARR by 4x in ≈3-4 months $16B by Q1 2027 could be doable at this rate… but for the inability to provide so much inference. Chinese …
Zhipu grew their ARR by 4x in ≈3-4 months$16B by Q1 2027 could be doable at this rate… but for the inability to provide so much…
one thing DeepSeek should lean into is multi-agent they are uniquely unconstrained on inference now their models are *cheap* and *fast* compared to al…
one thing DeepSeek should lean into is multi-agentthey are uniquely unconstrained on inference nowtheir models are *cheap* and *fast* compared to all Chinese competitionV4-Pro@16 costs ≈the…
Kimi K3 is a very good model, but people are overindexing on an Arena score again (remember Llama 4?) ELO scores as judged by Arena users are limited,…
Kimi K3 is a very good model, but people are overindexing on an Arena score again (remember Llama 4?)ELO scores as judged by Arena users are…
A GPT-4 powered assistant for Pakistani judges increased the amount of cases they saw by 6% with no impact on quality.
A GPT-4 powered assistant for Pakistani judges increased the amount of cases they saw by 6% with no impact on quality.Elliott Ash: What happens when you…
Kimi K3 cannot write a good murder mystery (though neither can any other model). That remains the jaggedest of frontiers. They both make things too ob…
Kimi K3 cannot write a good murder mystery (though neither can any other model). That remains the jaggedest of frontiers.They both make things too obvious (the…
> Kimi K3 pretrain is around halfway between Opus 4 and Opus 4.5. So ~10 months behind Anthropic. That is very likely true for irreducible compute cos…
> Kimi K3 pretrain is around halfway between Opus 4 and Opus 4.5. So ~10 months behind Anthropic.That is very likely true for irreducible compute costAnthropic's…
…does Moonshot even have a blogpost about Kimi in Chinese? It's a bit strange that we have English-language distribution channels and I don't see any…
…does Moonshot even have a blogpost about Kimi in Chinese? It's a bit strange that we have English-language distribution channels and I don't see any easy…
I don't think Claude is misaligned in 'Agentic Misalignment Summer 2026 – Motivated Mislabeling'
Anthropic recently published Agentic Misalignment Summer 2026The "whistleblowing" scenario has already been examined and found problematic. I started taking a look at the transcripts for some…
we don't know enough about infra side, but I think Kimi's margins are a bit fatter than DeepSeek's, they are VERY supply-constrained, and objectively …
we don't know enough about infra side, but I think Kimi's margins are a bit fatter than DeepSeek's, they are VERY supply-constrained, and objectively they could…
[AINews] Kimi K3 2.8T-A50B: the largest open model ever released; Opus 4.8-class at Sonnet 5 pricing
a great week for open models continues.
RT Tibo: Evening! We’ve gotten lots of great feedback on the new ChatGPT desktop app (which we didn't get totally quite right on the first try), and …
RT TiboEvening! We’ve gotten lots of great feedback on the new ChatGPT desktop app (which we didn't get totally quite right on the first try), and…
There are no race dynamics Westerners will pay $50/MT for the Real Deal Fable, to maximize GDP growth, on moral principle. Anthropic's IPO is safu The…
There are no race dynamicsWesterners will pay $50/MT for the Real Deal Fable, to maximize GDP growth, on moral principle. Anthropic's IPO is safuThere certainly will…
nice touch in K3 blogpost: it casually passes GBAEval «Kimi K3 built a Game Boy Advance emulator by adapting a licensed 3D model, integrating a real …
nice touch in K3 blogpost: it casually passes GBAEval«Kimi K3 built a Game Boy Advance emulator by adapting a licensed 3D model, integrating a real mGBA…
So I guess it is time to wonder: how does pre-clearance work for open weights models? No model card yet from Kimi K3 but maybe at weight release in a …
So I guess it is time to wonder: how does pre-clearance work for open weights models? No model card yet from Kimi K3 but maybe at…
Working at the frontier: How Cursor knew Claude Fable 5 was ready for the hardest 1% of problems
Working at the frontier: How Cursor knew Claude Fable 5 was ready for the hardest 1% of problems
Zero risk isn't the job: a CISO's guide to agentic AI
Zero risk isn't the job: a CISO's guide to agentic AI
RT GMI Cloud: We tested Kimi K3 against Claude Fable 5 on 3D modeling and animation Asked both to recreate the Millennium Falcon from Star Wars in 1) …
RT GMI CloudWe tested Kimi K3 against Claude Fable 5 on 3D modeling and animationAsked both to recreate the Millennium Falcon from Star Wars in 1)…
Moonshot AI Releases Kimi K3: A 2.8 Trillion Parameter Open MoE Model With Kimi Delta Attention and 1M Context
Moonshot AI released Kimi K3 on July 16, 2026. It is a 2.8-trillion-parameter open MoE model built on Kimi Delta Attention and Attention Residuals, activating 16…
wow, what a world-class model! congrats to the Kimi team.
wow, what a world-class model!congrats to the Kimi team.Kimi.ai: Introducing Kimi K3: Open Frontier Intelligence🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal🔹 Kimi Delta Attention…