Загрузка видео...

Не удалось загрузить видео

На главную

🚨Gemini 3.5 Pro leak just dropped and the frontend game is actually different. Targeting July 17 release with: • 2M token context window (this might be the real game changer) • New “Deep Think” reasoning mode • Significantly better UI & frontend generation • Cleaner design taste + stronger...

intheworldofai's profile picture

WorldofAI

17,934 subscribers

91,230 просмотров • 1 месяц назад •via X (Twitter)

Новости и политикаОбразованиеНаука и технологии
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

CHINA JUST DROPPED AN AI CODING MODEL WITH A 1M CONTEXT WINDOW. And I connected it to Claude Code to see what it could actually do. Meet GLM-X Preview On paper, a few things immediately stood out: → 1M context window → Agentic coding capabilities → Works inside Claude Code → Designed for large-scale coding and reasoning workflows But specs don't matter much if the model can't deliver in practice. So I gave it a real-world task. THE TEST One prompt: > Build a modern AI lead generation dashboard using React and Tailwind CSS. Requirements: → Dark mode → Analytics dashboard → Lead table → Email outreach section → Responsive design → Production-ready component structure Instead of generating a few snippets, it planned the architecture, generated the dashboard components, created the Tailwind configuration, and walked through the implementation requirements. What impressed me most wasn't the code itself. It was how well it maintained context throughout the workflow. That's where a 1M context window starts becoming useful. Less time re-explaining requirements. Less context loss. More room for complex projects. The AI coding race is getting very interesting. And it's no longer just GPT, Claude, and Gemini competing for attention. Results from my test below 👇
0:33

CHINA JUST DROPPED AN AI CODING MODEL WITH A 1M CONTEXT WINDOW. And I connected it to Claude Code to see what it could actually do. Meet GLM-X Preview On paper, a few things immediately stood out: → 1M context window → Agentic coding capabilities → Works inside Claude Code → Designed for large-scale coding and reasoning workflows But specs don't matter much if the model can't deliver in practice. So I gave it a real-world task. THE TEST One prompt: > Build a modern AI lead generation dashboard using React and Tailwind CSS. Requirements: → Dark mode → Analytics dashboard → Lead table → Email outreach section → Responsive design → Production-ready component structure Instead of generating a few snippets, it planned the architecture, generated the dashboard components, created the Tailwind configuration, and walked through the implementation requirements. What impressed me most wasn't the code itself. It was how well it maintained context throughout the workflow. That's where a 1M context window starts becoming useful. Less time re-explaining requirements. Less context loss. More room for complex projects. The AI coding race is getting very interesting. And it's no longer just GPT, Claude, and Gemini competing for attention. Results from my test below 👇

Md Riyazuddin

31,199 просмотров • 2 месяцев назад

Welcome to Shipmas Day 3 part 1… the big one. 🚢🎁 Today, we remove the ceiling on what you can generate. Starting today, our new default agent is Thinking with Gemini 3 Pro. This agent uses advanced reasoning to "think" through your design before writing a single line of code. The community has been calling Gemini 3 "too good at frontend" and "unmatched in visual reasoning." We agree. It handles complex dashboards, intricate DOM manipulation, and pixel-perfect CSS better than anything we've tested. All designs are backed by code and exportable to your favorite coding agent. The video below shows some single prompt examples to inspire you. Sound on for a preview of the new Stitch theme song 🔊
0:43

Welcome to Shipmas Day 3 part 1… the big one. 🚢🎁 Today, we remove the ceiling on what you can generate. Starting today, our new default agent is Thinking with Gemini 3 Pro. This agent uses advanced reasoning to "think" through your design before writing a single line of code. The community has been calling Gemini 3 "too good at frontend" and "unmatched in visual reasoning." We agree. It handles complex dashboards, intricate DOM manipulation, and pixel-perfect CSS better than anything we've tested. All designs are backed by code and exportable to your favorite coding agent. The video below shows some single prompt examples to inspire you. Sound on for a preview of the new Stitch theme song 🔊

Stitch by Google

192,446 просмотров • 8 месяцев назад

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!
2:57

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!

Andrew Ng

201,127 просмотров • 1 год назад

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!
0:20

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!

Jim Fan

278,517 просмотров • 2 лет назад

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)
0:40

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)

kwindla

40,593 просмотров • 3 месяцев назад

Introducing SuperDesign.Dev The first open-source Design Agent that lives inside Cursor, Windsurf, or any IDE: 👉 Instantly spin up 10 different design in parallel 👉 Explore, fork, and iterate 10x faster 👉 Create new or iterate on your existing UI — all fully local JackJack and I shipped this as a quick weekend experiment to validate a hunch: design is the killer use case for parallel agents. Currently at V0.005 — a simple customization of the Claude Code SDK + static HTML generation. (We might be the first product powered by the Anthropic Claude Code SDK!) More coming soon: web app, mobile app, SVG generation, and more. GitHub: Try it out: ----- Upvote on Hacker News ❤️:
0:55

Introducing SuperDesign.Dev The first open-source Design Agent that lives inside Cursor, Windsurf, or any IDE: 👉 Instantly spin up 10 different design in parallel 👉 Explore, fork, and iterate 10x faster 👉 Create new or iterate on your existing UI — all fully local JackJack and I shipped this as a quick weekend experiment to validate a hunch: design is the killer use case for parallel agents. Currently at V0.005 — a simple customization of the Claude Code SDK + static HTML generation. (We might be the first product powered by the Anthropic Claude Code SDK!) More coming soon: web app, mobile app, SVG generation, and more. GitHub: Try it out: ----- Upvote on Hacker News ❤️:

Jason Zhou

49,435 просмотров • 1 год назад

🎉 Congrats to MiniMax (official) on releasing MiniMax M3! Frontier coding and agentic capabilities, native image and video input, computer use, and a 1M-token context window, all in a single open model. At the heart of M3 is MSA, a new sparse attention architecture: instead of attending densely over the full KV cache, each query scores 128-token KV blocks and runs attention only over the top blocks. That is what makes 1M-token context practical to serve. M3 runs in vLLM with day-0 support, verified on NVIDIA and AMD hardware: ✨ MSA sparse attention with dedicated prefill and decode kernels ✨ 1M-token context serving with prefix caching and chunked prefill ✨ BF16 and MXFP8 checkpoints, with MoE backends for both Hopper and Blackwell ✨ Native multimodal input (image + video) ✨ Tool calling, reasoning parsing, and thinking-mode control for agent workloads Day-0 support like this is a true team effort. Grateful to the teams at MiniMax (official), NVIDIA AI, AI at AMD, and Inferact, and to the vLLM community for making it happen. 🙏 Deep dive into the implementation, kernel work, and deployment recipes: 🔗
0:21

🎉 Congrats to MiniMax (official) on releasing MiniMax M3! Frontier coding and agentic capabilities, native image and video input, computer use, and a 1M-token context window, all in a single open model. At the heart of M3 is MSA, a new sparse attention architecture: instead of attending densely over the full KV cache, each query scores 128-token KV blocks and runs attention only over the top blocks. That is what makes 1M-token context practical to serve. M3 runs in vLLM with day-0 support, verified on NVIDIA and AMD hardware: ✨ MSA sparse attention with dedicated prefill and decode kernels ✨ 1M-token context serving with prefix caching and chunked prefill ✨ BF16 and MXFP8 checkpoints, with MoE backends for both Hopper and Blackwell ✨ Native multimodal input (image + video) ✨ Tool calling, reasoning parsing, and thinking-mode control for agent workloads Day-0 support like this is a true team effort. Grateful to the teams at MiniMax (official), NVIDIA AI, AI at AMD, and Inferact, and to the vLLM community for making it happen. 🙏 Deep dive into the implementation, kernel work, and deployment recipes: 🔗

vLLM

40,306 просмотров • 2 месяцев назад

A deep dive into $SEXY and why I believe this is one of the most undervalued gamefi projects in the ecosystem: A simple beginning of a frictionless gaming experience fitting itself into a mobile game market which produced $123B in revenue in 2023 and an expected groth into $190B into 2030. The current game features a simple UI with strong incentives such a $1M prize pool in tokens lasting from June to September for early users pre-layer 3 buildout. Though simple in design, the transaction volume is telling with 300k user txns within 3 days of opening. The game is very much similar to the old micro transaction mini games such as clash of clans. Daily activities to keep users engaged and retained especially given the incentivization program. My thesis around this is the growth reward potential. Currently as the game sits in a more private beta (only users with codes can participate), we are at the simple version of the ecosystem. As expansion happens and there is more use items, more mini-games, more access, and more use for $SEXY, the value will grow. I imagine the R/R of investing in a game like clash of clans early on, imagine being able to invest in the value of gems in the early days and the value of these changed with user demand. Clash of clans did $360m in revenue in 2023. This model would recreate that volume all flowing into the token (which would be similar to gems). Thus, the thesis is simple: $SEXY ecosystem grows, more usage for active players to use the token within the mini games, more addictive tendencies lead to users wanting to gain more loot, more referrals to beat their friends in the ecosystem all leads to a flywheel of growth; especially, as users are token holders and their dominant strategy is to play the game for incentives and to invite their friends to grow the token value and ecosystem. Personally, I will be staking my $SEXY, collecting loot (imagine the loot can be tradable like early Runescape party hats), building my ranking for the incentives, and encouraging the ecosystem growth as a user in the flywheel. Thus, I believe as time passes, the value of the token continually increases under the assumption the ecosystem is growing with it. Speculation transitions from I think future buyers will find this token a good investment into I think players will buy this token to play the games within the ecosystem. Disclaimer: I am an early investor in ETHXY and so is WWVentures. All thesis generation may be biased.
0:42

A deep dive into $SEXY and why I believe this is one of the most undervalued gamefi projects in the ecosystem: A simple beginning of a frictionless gaming experience fitting itself into a mobile game market which produced $123B in revenue in 2023 and an expected groth into $190B into 2030. The current game features a simple UI with strong incentives such a $1M prize pool in tokens lasting from June to September for early users pre-layer 3 buildout. Though simple in design, the transaction volume is telling with 300k user txns within 3 days of opening. The game is very much similar to the old micro transaction mini games such as clash of clans. Daily activities to keep users engaged and retained especially given the incentivization program. My thesis around this is the growth reward potential. Currently as the game sits in a more private beta (only users with codes can participate), we are at the simple version of the ecosystem. As expansion happens and there is more use items, more mini-games, more access, and more use for $SEXY, the value will grow. I imagine the R/R of investing in a game like clash of clans early on, imagine being able to invest in the value of gems in the early days and the value of these changed with user demand. Clash of clans did $360m in revenue in 2023. This model would recreate that volume all flowing into the token (which would be similar to gems). Thus, the thesis is simple: $SEXY ecosystem grows, more usage for active players to use the token within the mini games, more addictive tendencies lead to users wanting to gain more loot, more referrals to beat their friends in the ecosystem all leads to a flywheel of growth; especially, as users are token holders and their dominant strategy is to play the game for incentives and to invite their friends to grow the token value and ecosystem. Personally, I will be staking my $SEXY, collecting loot (imagine the loot can be tradable like early Runescape party hats), building my ranking for the incentives, and encouraging the ecosystem growth as a user in the flywheel. Thus, I believe as time passes, the value of the token continually increases under the assumption the ecosystem is growing with it. Speculation transitions from I think future buyers will find this token a good investment into I think players will buy this token to play the games within the ecosystem. Disclaimer: I am an early investor in ETHXY and so is WWVentures. All thesis generation may be biased.

Crypto Max

23,392 просмотров • 2 лет назад

Day 13 Update - Aida the browser AI assistant. Aida can now think and execute plans - watch it scrape the current tab for all video titles on Rahul Mathur's Breakdown channel and save it as a txt file. Note that none of this code is written in advance or just a tool call, it writes code in realtime and dynamically executes it in the context of the browser. - Worked on the DOM Engine which provides better context to the assistant for dynamic code execution. - Wasn't satisfied with the design in the last update, tried to make it better. Whats next? Getting it to the hands of users. I've been gate-keeping this for too long. Made an accountability bet with Yogini Bende today to release it in a day.
0:44

Day 13 Update - Aida the browser AI assistant. Aida can now think and execute plans - watch it scrape the current tab for all video titles on Rahul Mathur's Breakdown channel and save it as a txt file. Note that none of this code is written in advance or just a tool call, it writes code in realtime and dynamically executes it in the context of the browser. - Worked on the DOM Engine which provides better context to the assistant for dynamic code execution. - Wasn't satisfied with the design in the last update, tried to make it better. Whats next? Getting it to the hands of users. I've been gate-keeping this for too long. Made an accountability bet with Yogini Bende today to release it in a day.

Nakshatra Saxena

10,971 просмотров • 1 год назад

Anthropic just changed how they think about Claude 5. Not with a new model. Not with a benchmark. With a completely different philosophy for building AI systems. Most people will miss it. They're still trying to write better prompts. Anthropic is optimizing something else entirely: Context. Here's what every AI builder should learn from it: 1. Stop telling AI exactly what to do. Start telling it what success looks like. Older models needed rigid instructions. Newer models perform better when you define the objective and let them make the decisions. The goal is no longer more control. It's more clarity. 2. Context is a budget, not a storage unit. Every extra sentence competes for the model's attention. A massive context file doesn't make AI smarter. It often makes reasoning worse. The best systems don't load everything. They load only what's relevant. 3. Great tools beat great prompts. Most people spend hours tweaking prompt wording. Anthropic is investing in better interfaces instead. Clear parameters. Structured inputs. Well-designed tools. If the interface removes ambiguity, the model makes better decisions before it even starts reasoning. 4. Show reality instead of describing it. Want a specific coding style? Provide the codebase. Want a certain design language? Share the mockup. Want consistent outputs? Give it tests to satisfy. Concrete references consistently outperform long written instructions. 5. Deliver information when it's needed. Not everything belongs in the initial context. Documentation. Style guides. Verification. Reviews. Load them only when they're relevant. Think of context like RAM, not a hard drive. 6. The real advantage is system design. Prompt engineering isn't disappearing. It's becoming one small part of a much bigger stack. Memory. Retrieval. Context architecture. Tool design. Evaluation. Every improvement compounds. The biggest takeaway from Anthropic's update? The next generation of AI won't be won by the people writing the longest prompts. It'll be won by the people building the smartest systems. Same models. Completely different results.
0:30

Anthropic just changed how they think about Claude 5. Not with a new model. Not with a benchmark. With a completely different philosophy for building AI systems. Most people will miss it. They're still trying to write better prompts. Anthropic is optimizing something else entirely: Context. Here's what every AI builder should learn from it: 1. Stop telling AI exactly what to do. Start telling it what success looks like. Older models needed rigid instructions. Newer models perform better when you define the objective and let them make the decisions. The goal is no longer more control. It's more clarity. 2. Context is a budget, not a storage unit. Every extra sentence competes for the model's attention. A massive context file doesn't make AI smarter. It often makes reasoning worse. The best systems don't load everything. They load only what's relevant. 3. Great tools beat great prompts. Most people spend hours tweaking prompt wording. Anthropic is investing in better interfaces instead. Clear parameters. Structured inputs. Well-designed tools. If the interface removes ambiguity, the model makes better decisions before it even starts reasoning. 4. Show reality instead of describing it. Want a specific coding style? Provide the codebase. Want a certain design language? Share the mockup. Want consistent outputs? Give it tests to satisfy. Concrete references consistently outperform long written instructions. 5. Deliver information when it's needed. Not everything belongs in the initial context. Documentation. Style guides. Verification. Reviews. Load them only when they're relevant. Think of context like RAM, not a hard drive. 6. The real advantage is system design. Prompt engineering isn't disappearing. It's becoming one small part of a much bigger stack. Memory. Retrieval. Context architecture. Tool design. Evaluation. Every improvement compounds. The biggest takeaway from Anthropic's update? The next generation of AI won't be won by the people writing the longest prompts. It'll be won by the people building the smartest systems. Same models. Completely different results.

Evan Luthra

38,751 просмотров • 1 месяц назад

Introducing NEO: The first Autonomous Machine Learning Engineer. It works like a full-stack ML engineer that never sleeps: handling data exploration, feature engineering, training, tuning, deployment, and monitoring, end to end. Powered by 11 specialized agents, NEO runs autonomously, saving ML engineers thousands of hours and making them 10x faster. Benchmarks: Tested on 75 Kaggle competitions, NEO scored a medal in 34.2% of them, significantly outperforming Microsoft’s RD Agent (22.4%) on OpenAI's MLE Bench. This sets a new state of the art for autonomous ML systems. NEO runs on a novel multi-agent orchestrator, powered by a multi-step reasoning engine, context transfer protocol, and agent memory, built to solve complex workflows end to end. And with human-in-the-loop mode, you can guide, inspect, and override any step. You're always in command. NEO is built for real world workflows and ready for production. NEO is here to make every ML engineer truly superhuman. Watch NEO in action:
3:51

Introducing NEO: The first Autonomous Machine Learning Engineer. It works like a full-stack ML engineer that never sleeps: handling data exploration, feature engineering, training, tuning, deployment, and monitoring, end to end. Powered by 11 specialized agents, NEO runs autonomously, saving ML engineers thousands of hours and making them 10x faster. Benchmarks: Tested on 75 Kaggle competitions, NEO scored a medal in 34.2% of them, significantly outperforming Microsoft’s RD Agent (22.4%) on OpenAI's MLE Bench. This sets a new state of the art for autonomous ML systems. NEO runs on a novel multi-agent orchestrator, powered by a multi-step reasoning engine, context transfer protocol, and agent memory, built to solve complex workflows end to end. And with human-in-the-loop mode, you can guide, inspect, and override any step. You're always in command. NEO is built for real world workflows and ready for production. NEO is here to make every ML engineer truly superhuman. Watch NEO in action:

Neo AI

860,919 просмотров • 1 год назад

MiniMax-M2.1 running fully local in AWQ-4Bit with full context window (170 GB VRAM w full context) - 1000~ to 16,000~ tps prefill - 100~ tps generation speeds - Opencode It’s doing real work, updating my blog with little steering or specificity. The problem with local LLMs is that they require too much steering, this means baby sitting which I don’t have the time to do MiniMax cracked the cost, intelligence, and speed challenge, I would say this is a top tier model. I run frontier models like Gemini and it just fails to call tools, in this year lol… ——————— I think glm-4.?-air is needed still. We need a viable model at each hardware entry point, a Mac M1 Ultra 192GB? is relatively cheap 5k to be able to run this model at 40 tps is a huge societal unlock. Smaller models can be good but size matters :p
3:01

MiniMax-M2.1 running fully local in AWQ-4Bit with full context window (170 GB VRAM w full context) - 1000~ to 16,000~ tps prefill - 100~ tps generation speeds - Opencode It’s doing real work, updating my blog with little steering or specificity. The problem with local LLMs is that they require too much steering, this means baby sitting which I don’t have the time to do MiniMax cracked the cost, intelligence, and speed challenge, I would say this is a top tier model. I run frontier models like Gemini and it just fails to call tools, in this year lol… ——————— I think glm-4.?-air is needed still. We need a viable model at each hardware entry point, a Mac M1 Ultra 192GB? is relatively cheap 5k to be able to run this model at 40 tps is a huge societal unlock. Smaller models can be good but size matters :p

0xSero

23,836 просмотров • 8 месяцев назад

Organizing a wedding isn't difficult because of one isolated choice. Learn more: Try it: It becomes tough because variables constantly shift. Guest lists, venues, schedules — every single decision alters what happens next. This dynamic challenge is exactly why dots studio's dots3-note preview tackles long-horizon tasks. Equipped with 280B total parameters, 16B active parameters, 512K context window, and multimodal capabilities, it is engineered to process complex, real-world problems. The key is TEMPO. A massive context window alone doesn't guarantee success. While traditional agents just plan, act, and wait for a final reward, TEMPO helps agents stay on track by evaluating and adjusting mid-process. The Actor executes the steps, while the Critic actively reviews progress and guides the next move — allowing the agent to correct its own mistakes before the task concludes. It also combines: 🌱 Learning in deployment — adapting to unfamiliar environments through continuous interaction and real-time feedback. 🌍 Multimodal understanding — linking text, vision, and audio to navigate tangible, everyday challenges. This represents the core vision of rednote: AI that genuinely adapts to real life. From managing extended workflows to proving its deep logic with IMO gold medal-level reasoning, dots3-note preview outlines a practical new path for AI agents.
1:03

Organizing a wedding isn't difficult because of one isolated choice. Learn more: Try it: It becomes tough because variables constantly shift. Guest lists, venues, schedules — every single decision alters what happens next. This dynamic challenge is exactly why dots studio's dots3-note preview tackles long-horizon tasks. Equipped with 280B total parameters, 16B active parameters, 512K context window, and multimodal capabilities, it is engineered to process complex, real-world problems. The key is TEMPO. A massive context window alone doesn't guarantee success. While traditional agents just plan, act, and wait for a final reward, TEMPO helps agents stay on track by evaluating and adjusting mid-process. The Actor executes the steps, while the Critic actively reviews progress and guides the next move — allowing the agent to correct its own mistakes before the task concludes. It also combines: 🌱 Learning in deployment — adapting to unfamiliar environments through continuous interaction and real-time feedback. 🌍 Multimodal understanding — linking text, vision, and audio to navigate tangible, everyday challenges. This represents the core vision of rednote: AI that genuinely adapts to real life. From managing extended workflows to proving its deep logic with IMO gold medal-level reasoning, dots3-note preview outlines a practical new path for AI agents.

Anwar Maqsood| Ai

20,196 просмотров • 4 дней назад

🚀 Exciting times for MATR1X FIRE We’re thrilled to announce our Token Generation Event (TGE) launching on OKX, BingX, and more exchanges. This marks a major milestone and the start of a promising journey. We’re grateful for your support as we move forward together! 🔹 Introduction to Matrix: Matrix is setting new standards in the gaming industry with its innovative approach and strong team commitment. 🔹 Positive Insights on $MAX: The $MAX token is poised for significant growth, especially with its current undervaluation and upcoming developments. 🔹 Key Points: The core focus is on the coin price, expected to be in the $1-2 range. Maintaining this price is crucial for the long-term success of the game. The team has a live game public blockchain on testnet, which is currently undervalued. Let’s make this journey unforgettable! 🚀🔥
1:27

🚀 Exciting times for MATR1X FIRE We’re thrilled to announce our Token Generation Event (TGE) launching on OKX, BingX, and more exchanges. This marks a major milestone and the start of a promising journey. We’re grateful for your support as we move forward together! 🔹 Introduction to Matrix: Matrix is setting new standards in the gaming industry with its innovative approach and strong team commitment. 🔹 Positive Insights on $MAX: The $MAX token is poised for significant growth, especially with its current undervaluation and upcoming developments. 🔹 Key Points: The core focus is on the coin price, expected to be in the $1-2 range. Maintaining this price is crucial for the long-term success of the game. The team has a live game public blockchain on testnet, which is currently undervalued. Let’s make this journey unforgettable! 🚀🔥

Mr|Mamba

12,454 просмотров • 2 лет назад

Introducing NEO: The first Agentic Machine Learning Engineer. It works like a full-stack ML engineer that never sleeps: handling data exploration, feature engineering, training, tuning, deployment, and monitoring, end to end. Powered by 11 specialized agents, NEO runs autonomously, saving ML engineers thousands of hours and making them 10x faster. Benchmarks: Tested on 75 Kaggle competitions, NEO scored a medal in 34% of them, significantly outperforming Microsoft’s RD Agent (22.4%) on the MLE Bench. This sets a new state of the art for autonomous ML systems. NEO runs on a novel multi-agent orchestrator, powered by a multi-step reasoning engine, context transfer protocol, and agent memory, built to solve complex workflows end to end. And with human-in-the-loop mode, you can guide, inspect, and override any step. You're always in command. NEO is built for real world workflows and ready for production. NEO is here to make every ML engineer truly superhuman. Watch NEO in action:
3:51

Introducing NEO: The first Agentic Machine Learning Engineer. It works like a full-stack ML engineer that never sleeps: handling data exploration, feature engineering, training, tuning, deployment, and monitoring, end to end. Powered by 11 specialized agents, NEO runs autonomously, saving ML engineers thousands of hours and making them 10x faster. Benchmarks: Tested on 75 Kaggle competitions, NEO scored a medal in 34% of them, significantly outperforming Microsoft’s RD Agent (22.4%) on the MLE Bench. This sets a new state of the art for autonomous ML systems. NEO runs on a novel multi-agent orchestrator, powered by a multi-step reasoning engine, context transfer protocol, and agent memory, built to solve complex workflows end to end. And with human-in-the-loop mode, you can guide, inspect, and override any step. You're always in command. NEO is built for real world workflows and ready for production. NEO is here to make every ML engineer truly superhuman. Watch NEO in action:

Neo AI

19,571,775 просмотров • 1 год назад

🚨 ALIBABA JUST OPEN-SOURCED THE MOST POWERFUL MODEL IN THE WORLD - Qwen3.5 plus And it changes the AI race forever. This is not another “bigger model” release. This is a complete architectural reset. 397B parameters → only 17B active at inference, Qwen3.5 plus delivers performance on par with GPT-5.2 and Gemini-3-Pro. A new Hybrid Sparse MoE + Linear Attention architecture + Multi-Token Prediction = near-instant long-context responses and massive speed gains. Built as a true native multimodal model, it doesn’t just chat it builds. With a single prompt: “Use the React framework to create a visually impressive Mapbox demo featuring locations in Beijing and Shanghai” it generates a live interactive 3D map UI, structured components, real layouts, and production-ready frontend logic. This is real agentic, multimodal coding not a static code block. MMLU-Pro: 87.8 GPQA: 88.4 IFBench: 76.5 ⚡ up to 8× faster inference And it’s fully open-sourced under Apache-2.0. No locked APIs. No restricted tiers. 201 languages supported bringing frontier AI to developers, startups, and researchers worldwide. The race is no longer about the biggest model. It’s about the most usable intelligence. Open source just entered the top tier.
1:50

🚨 ALIBABA JUST OPEN-SOURCED THE MOST POWERFUL MODEL IN THE WORLD - Qwen3.5 plus And it changes the AI race forever. This is not another “bigger model” release. This is a complete architectural reset. 397B parameters → only 17B active at inference, Qwen3.5 plus delivers performance on par with GPT-5.2 and Gemini-3-Pro. A new Hybrid Sparse MoE + Linear Attention architecture + Multi-Token Prediction = near-instant long-context responses and massive speed gains. Built as a true native multimodal model, it doesn’t just chat it builds. With a single prompt: “Use the React framework to create a visually impressive Mapbox demo featuring locations in Beijing and Shanghai” it generates a live interactive 3D map UI, structured components, real layouts, and production-ready frontend logic. This is real agentic, multimodal coding not a static code block. MMLU-Pro: 87.8 GPQA: 88.4 IFBench: 76.5 ⚡ up to 8× faster inference And it’s fully open-sourced under Apache-2.0. No locked APIs. No restricted tiers. 201 languages supported bringing frontier AI to developers, startups, and researchers worldwide. The race is no longer about the biggest model. It’s about the most usable intelligence. Open source just entered the top tier.

RAVI KUMAR SAHU

380,123 просмотров • 6 месяцев назад

We are witnessing a fundamental shift in how software is built. Over the decades, UI was static and predictable. But AI is changing the fundamental nature of the interface. We are moving away from fixed UI and toward a new standard: Generative UI. Generative UI lets agents participate in the UI, not just the conversation. It’s a dynamic participant in the user experience. Depending on the complexity of your app, Generative UI generally falls into three categories: 1. Static (AG-UI): UI is chosen from a fixed set of hand-built components 2. Open-Ended (MCP Apps): agent sends arbitrary content like raw HTML or iframes for total flexibility. 3. Declarative (A2UI): agent and frontend communicate via a structured UI spec for dynamic widgets and forms. The AG-UI Protocol by CopilotKit is the real breakthrough. AG-UI acts as the universal transport layer. It provides the integrations and middleware for every Gen-UI spec, making it trivial to bring dynamic components into your existing app. CopilotKit provides the go-to ecosystem for building agentic applications and is the React client of AG-UI. It is designed to be deeply integrated, allowing developers to bring their own frameworks or protocols to the table. Play around with different types of Generative UI here → Check out the video attached to see Generative UI in action. Appreciate the CopilotKit team for collaborating with me on this deep dive.
0:33

We are witnessing a fundamental shift in how software is built. Over the decades, UI was static and predictable. But AI is changing the fundamental nature of the interface. We are moving away from fixed UI and toward a new standard: Generative UI. Generative UI lets agents participate in the UI, not just the conversation. It’s a dynamic participant in the user experience. Depending on the complexity of your app, Generative UI generally falls into three categories: 1. Static (AG-UI): UI is chosen from a fixed set of hand-built components 2. Open-Ended (MCP Apps): agent sends arbitrary content like raw HTML or iframes for total flexibility. 3. Declarative (A2UI): agent and frontend communicate via a structured UI spec for dynamic widgets and forms. The AG-UI Protocol by CopilotKit is the real breakthrough. AG-UI acts as the universal transport layer. It provides the integrations and middleware for every Gen-UI spec, making it trivial to bring dynamic components into your existing app. CopilotKit provides the go-to ecosystem for building agentic applications and is the React client of AG-UI. It is designed to be deeply integrated, allowing developers to bring their own frameworks or protocols to the table. Play around with different types of Generative UI here → Check out the video attached to see Generative UI in action. Appreciate the CopilotKit team for collaborating with me on this deep dive.

Pratham

33,920 просмотров • 7 месяцев назад

Anthropic Managed Agents team: "Fable 5 is our best model for running self-improving agent systems. Add /loops, dynamic workflows, dreaming and you are unstoppable" in this video he breaks down exactly how most people are using Claude: - the 34% you lose to CLAUDE.md before typing a word - one agent researching. one building. one reviewing. one orchestrating - the architecture that separates hobbyists from real builders - the 3 properties every agent team needs to actually survive if you've been using Claude for more than a month and never left the chat window, you've been using one agent when you could be running a team of them instead of another show tonight, watch this make sure to bookmark it before it gets lost in your feed the guide is in the article below
13:23

Anthropic Managed Agents team: "Fable 5 is our best model for running self-improving agent systems. Add /loops, dynamic workflows, dreaming and you are unstoppable" in this video he breaks down exactly how most people are using Claude: - the 34% you lose to CLAUDE.md before typing a word - one agent researching. one building. one reviewing. one orchestrating - the architecture that separates hobbyists from real builders - the 3 properties every agent team needs to actually survive if you've been using Claude for more than a month and never left the chat window, you've been using one agent when you could be running a team of them instead of another show tonight, watch this make sure to bookmark it before it gets lost in your feed the guide is in the article below

Khairallah AL-Awady

21,533 просмотров • 2 месяцев назад

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!
0:39

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!

Milk Road AI

128,678 просмотров • 1 месяц назад