正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

In practice, this advantage is even more apparent than in benchmarks. For example: a task to move 151 Linear issues and verify each took 312 tool calls in the past. With direct calls, that's about 36 requests and 3.1M input tokens. Now, as a Script, it took 2 requests... show more

Factory

59,329 subscribers

47,085 次观看 • 6 天前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

7 条评论

Factory 的头像

Factory6 天前

The biggest change in Droid 1.0: programmatic tool calling is now the default. Instead of calling tools one at a time, Droid writes a program that is capable of running up to hundreds of tools per single model request.

Factory 的头像

Factory6 天前

On Terminal-Bench 4, Droid outperforms popular models’ native harnesses, with up to 20% better efficiency.

Factory 的头像

Factory6 天前

Introducing Droid 1.0: the new engine behind Factory’s agent. Built for complex projects and everyday development: >more accurate >more efficient >faster than ever before

Factory 的头像

Factory6 天前

We built Droid 1.0 on our core design principles and an internal Agent Factory that tests every change on every model family, which we outline in our blog. Read more about our methodology here:

gary IH fung 的头像

gary IH fung5 天前

how does Droid compare with Pi though?

SimonLab 的头像

SimonLab5 天前

LLMs already get the idea of writing a helper script for repetitive tasks. Great idea to embrace such a flow on a harness level; side effect: agents probably do more targeted mid-job reasoning bursts. Testing now.

climate alarmist 的头像

climate alarmist5 天前

Phenomenal

相关视频

Mistral, your "best US/European open weights model" made this goddamn garbage. I asked Large 4 for a Three.js FPS with good gunplay. 5.30M tokens across all requests, including 4.33M cached input. 98 tool calls, $1.00 through OpenRouter. Solo agent. It tested itself in a browser and called the game complete and fine. I tried it and it's fucking awful. You should include this demo in the announcement too.

Mistral, your "best US/European open weights model" made this goddamn garbage. I asked Large 4 for a Three.js FPS with good gunplay. 5.30M tokens across all requests, including 4.33M cached input. 98 tool calls, $1.00 through OpenRouter. Solo agent. It tested itself in a browser and called the game complete and fine. I tried it and it's fucking awful. You should include this demo in the announcement too.

Alexey Fateev

19,461 次观看 • 29 天前

beomgyu executing the move with even More vibration than the dancers, the precision in each move while executing each one to the fullest and maintaining the facial expressions needed for the performance he is such a star performer and dancer and i need everyone to talk about it

beomgyu executing the move with even More vibration than the dancers, the precision in each move while executing each one to the fullest and maintaining the facial expressions needed for the performance he is such a star performer and dancer and i need everyone to talk about it

🌼

28,483 次观看 • 10 个月前

Suns have a chance to run the clock out, and at a 5-point gap, Booker calls for Dunn to score. He does, and as the Pelicans look OK with ending it, Queen calls for the ball to go score (as he has in the past). Booker denies and goes for a swipe. Immediately calls for review LOL

Suns have a chance to run the clock out, and at a 5-point gap, Booker calls for Dunn to score. He does, and as the Pelicans look OK with ending it, Queen calls for the ball to go score (as he has in the past). Booker denies and goes for a swipe. Immediately calls for review LOL

Kellan Olson

1,087,152 次观看 • 10 个月前

Introducing Batch Bridge, powered by Relay A tool that allow users to bridge or swaps multiple tokens across Ethereum, Base and Arbitrum in a single batch transaction using EIP5792 and EIP7702 🔗 Tool Link : 🔗 Code : Batching very small value tokens may cause issues, so it is best to batch tokens worth at least $1. This rool uses EIP-5792 for batching when supported. Wallets that do not support wallet_sendCalls fall back to sequential transactions. For the true batch transaction experience, use ambire.eth wallet or OKX wallet, they fully support EIP-5792 batch calls, allowing all swaps and approvals to execute in a single transaction.

Introducing Batch Bridge, powered by Relay A tool that allow users to bridge or swaps multiple tokens across Ethereum, Base and Arbitrum in a single batch transaction using EIP5792 and EIP7702 🔗 Tool Link : 🔗 Code : Batching very small value tokens may cause issues, so it is best to batch tokens worth at least $1. This rool uses EIP-5792 for batching when supported. Wallets that do not support wallet_sendCalls fall back to sequential transactions. For the true batch transaction experience, use ambire.eth wallet or OKX wallet, they fully support EIP-5792 batch calls, allowing all swaps and approvals to execute in a single transaction.

Zun

60,118 次观看 • 10 个月前

NEWS: X Calls is coming, a meeting tool for everyone. It is X's own group video meeting product, a rival to Zoom and Google Meet, now in beta at What it offers: ~ Instant calls, scheduled calls or a call saved for later. ~ A built-in calendar to see all your events and calls in one place. ~ Shareable links and codes, with Google Calendar and Microsoft Teams integration. ~ A lobby with camera and mic checks before you join. ~ Reactions, custom backdrops, a whiteboard, screen sharing and live chat. ~ It works on web, Android and iOS, with links that open without signing in.

NEWS: X Calls is coming, a meeting tool for everyone. It is X's own group video meeting product, a rival to Zoom and Google Meet, now in beta at What it offers: ~ Instant calls, scheduled calls or a call saved for later. ~ A built-in calendar to see all your events and calls in one place. ~ Shareable links and codes, with Google Calendar and Microsoft Teams integration. ~ A lobby with camera and mic checks before you join. ~ Reactions, custom backdrops, a whiteboard, screen sharing and live chat. ~ It works on web, Android and iOS, with links that open without signing in.

Muskonomy

37,818 次观看 • 1 个月前

Redis built a cache that cuts LLM costs by 90%! Production LLM apps do not receive completely new questions every time. A customer-support assistant might receive all three of these: - "Can I get a refund after buying the monthly plan?" - "Is the monthly subscription refundable?" - "Can I cancel the plan and get my money back?" The wording is different, but the underlying question and its answer remain the same. Yet LLM apps process every version as a new request. They assemble the prompt, send it to the model, and generate an answer that may have already been generated. Prefix caching reduces part of these repeated calls. When requests begin with the same system prompt or context, the model can reuse the KV states already computed for that shared prefix. But the request still hits the LLM. The new tokens must be processed, and the complete answer must still be decoded. So even with a prefix-cache hit, there's another generation call involved. To solve this, instead of only caching computation inside the model, the application can cache the generated response outside it. When another question arrives, the system embeds it and compares it with previously answered questions. If it finds a sufficiently close match, it returns the stored response without invoking the LLM again. A cache hit removes the input tokens, output tokens, and decoding time associated with another LLM call. In practice, it is important to decide which questions can safely share an answer since a production setup needs well-tuned similarity thresholds, expiration policies, data isolation, and monitoring for incorrect matches. If you want to use this in practice, Redis already implements it as a managed service called Redis LangCache. Under the hood, it generates embeddings, searches previous responses, and returns a matching answer before another model call occurs. Redis also handles access scopes, custom filtering, TTL and eviction controls, and cache monitoring through Redis Cloud. I built an interface to compare it against direct LLM inference. The video below shows this in action, and I worked with Redis on this post to put this together. For the paraphrased question in my run, direct inference took 2.232 seconds and consumed 514 input tokens plus 250 output tokens. Redis returned the earlier response in 0.37 seconds with zero LLM input or output tokens. That was roughly 6x faster in this run. Redis reports API cost savings of up to 90% and cache-hit responses up to 15x faster. The actual result depends on how much safe repetition exists in the workload. You can try Redis LangCache here: If you want to dive deeper, I have already written a detailed breakdown of KV, prefix, prompt, and semantic caching in the article quoted below. This demo builds on the final technique and shows it running in practice. Read it below.

Redis built a cache that cuts LLM costs by 90%! Production LLM apps do not receive completely new questions every time. A customer-support assistant might receive all three of these: - "Can I get a refund after buying the monthly plan?" - "Is the monthly subscription refundable?" - "Can I cancel the plan and get my money back?" The wording is different, but the underlying question and its answer remain the same. Yet LLM apps process every version as a new request. They assemble the prompt, send it to the model, and generate an answer that may have already been generated. Prefix caching reduces part of these repeated calls. When requests begin with the same system prompt or context, the model can reuse the KV states already computed for that shared prefix. But the request still hits the LLM. The new tokens must be processed, and the complete answer must still be decoded. So even with a prefix-cache hit, there's another generation call involved. To solve this, instead of only caching computation inside the model, the application can cache the generated response outside it. When another question arrives, the system embeds it and compares it with previously answered questions. If it finds a sufficiently close match, it returns the stored response without invoking the LLM again. A cache hit removes the input tokens, output tokens, and decoding time associated with another LLM call. In practice, it is important to decide which questions can safely share an answer since a production setup needs well-tuned similarity thresholds, expiration policies, data isolation, and monitoring for incorrect matches. If you want to use this in practice, Redis already implements it as a managed service called Redis LangCache. Under the hood, it generates embeddings, searches previous responses, and returns a matching answer before another model call occurs. Redis also handles access scopes, custom filtering, TTL and eviction controls, and cache monitoring through Redis Cloud. I built an interface to compare it against direct LLM inference. The video below shows this in action, and I worked with Redis on this post to put this together. For the paraphrased question in my run, direct inference took 2.232 seconds and consumed 514 input tokens plus 250 output tokens. Redis returned the earlier response in 0.37 seconds with zero LLM input or output tokens. That was roughly 6x faster in this run. Redis reports API cost savings of up to 90% and cache-hit responses up to 15x faster. The actual result depends on how much safe repetition exists in the workload. You can try Redis LangCache here: If you want to dive deeper, I have already written a detailed breakdown of KV, prefix, prompt, and semantic caching in the article quoted below. This demo builds on the final technique and shows it running in practice. Read it below.

Avi Chawla

166,230 次观看 • 1 个月前

Claude Code used 3x fewer tokens with one change: - Before: 10.4M tokens · 10 errors · $9.21 cost - After: 3.7M tokens · 0 errors · $2.81 cost I used InsForge Skills + CLI as the backend context engineering layer for Claude Code (open-source and local). With Supabase, tables, RLS policies, auth providers, storage buckets, and edge functions were retrieved through separate calls. Each response entered the conversation history, so later calls carried more context. But InsForge returned the backend topology in one metadata call of roughly 500 tokens. InsForge repo: (don't forget to star it 🌟) Its instructions were also split into narrow skills for CLI operations, debugging, SDK usage, and integrations. Claude loaded only the instructions needed for the task. The CLI returned structured JSON with semantic exit codes as well. This made it clear whether a failure came from the operation itself or the code being executed, avoiding several retry loops. The token savings essentially came from replacing repeated backend discovery and debugging calls with one structured view of the backend. I wrote a detailed walkthrough of both builds, including where the extra calls, tokens, and errors came from. Read it below.

Claude Code used 3x fewer tokens with one change: - Before: 10.4M tokens · 10 errors · $9.21 cost - After: 3.7M tokens · 0 errors · $2.81 cost I used InsForge Skills + CLI as the backend context engineering layer for Claude Code (open-source and local). With Supabase, tables, RLS policies, auth providers, storage buckets, and edge functions were retrieved through separate calls. Each response entered the conversation history, so later calls carried more context. But InsForge returned the backend topology in one metadata call of roughly 500 tokens. InsForge repo: (don't forget to star it 🌟) Its instructions were also split into narrow skills for CLI operations, debugging, SDK usage, and integrations. Claude loaded only the instructions needed for the task. The CLI returned structured JSON with semantic exit codes as well. This made it clear whether a failure came from the operation itself or the code being executed, avoiding several retry loops. The token savings essentially came from replacing repeated backend discovery and debugging calls with one structured view of the backend. I wrote a detailed walkthrough of both builds, including where the extra calls, tokens, and errors came from. Read it below.

Avi Chawla

65,041 次观看 • 1 个月前

A viral tweet claimed that Meta is now responsible for nearly a third of Anthropic's $30B ARR, based on reporting that Meta used 60.2T tokens over 30 days. This doesn't actually pencil out. It assumes all of Meta's spend is on output tokens, which are much more expensive than input tokens, and which would be extremely unusual — according to OpenRouter data, about 98.9% of all Opus 4.6 tokens are input. The actual number is *maybe* closer to $136M a month, or $1.6B a year. Tyler Cosgrove goes through the math:

A viral tweet claimed that Meta is now responsible for nearly a third of Anthropic's $30B ARR, based on reporting that Meta used 60.2T tokens over 30 days. This doesn't actually pencil out. It assumes all of Meta's spend is on output tokens, which are much more expensive than input tokens, and which would be extremely unusual — according to OpenRouter data, about 98.9% of all Opus 4.6 tokens are input. The actual number is maybe closer to $136M a month, or $1.6B a year. Tyler Cosgrove goes through the math:

TBPN

254,894 次观看 • 7 个月前

Streamlining the UX in Gemini CLI ✨ We got a lot of feedback on how our UI was way too cluttered with noise and text... so we made changes! 🛠️ Compact tool calls: Tool calls for reading files, folders, searching text, etc are now a single line (no more tool boxes around everything!) 💭 Topics: The agent outputs a one line overview of the rationale and direction it is going. A topic can span several tool calls and makes it easy to see what the agent is working on at a glance and why.

Streamlining the UX in Gemini CLI ✨ We got a lot of feedback on how our UI was way too cluttered with noise and text... so we made changes! 🛠️ Compact tool calls: Tool calls for reading files, folders, searching text, etc are now a single line (no more tool boxes around everything!) 💭 Topics: The agent outputs a one line overview of the rationale and direction it is going. A topic can span several tool calls and makes it easy to see what the agent is working on at a glance and why.

Jack Wotherspoon

16,053 次观看 • 6 个月前

Bitcoin is how drug cartels operating in Vancouver move and launder money. The crime and safety issues in our city are a direct result of this. We need to make it harder for criminals to move money with Bitcoin–not heating pools with it.

Bitcoin is how drug cartels operating in Vancouver move and launder money. The crime and safety issues in our city are a direct result of this. We need to make it harder for criminals to move money with Bitcoin–not heating pools with it.

Kareem Allam

73,465 次观看 • 1 年前

Pi Agent vs OpenCode token usage A lot of people recommended Pi Agent so I decided to check Pi Agent took 1.1k tokens in first turn OpenCode took 11.5k Setup: 1) Trimmed OpenCode (from usual 30k first turn to 11.5k) - 0 MCPs - 2 lightweight plugins (opencode-env-protect and openslimedit) - 8k char AGENTS.md - 11585 deepseek-v4-flash input tokens - for $0.0016 2) Vanilla Pi - 0 MCPs - 0 system prompt - 8k char AGENTS.md - 1114 kimi-k2.6 input tokens - for $0.0008 I think using better models with capped tokens per turn can keep usage nearly the same as uncontrolled DeepSeek V4 Flash? The challenge now is finding the sweet spot. Can't cap tokens if quality drops. We'll see. Video below - OpenCode vs Pi side by side "say hi back" first turns test, with OpenCode Go usage for each

Pi Agent vs OpenCode token usage A lot of people recommended Pi Agent so I decided to check Pi Agent took 1.1k tokens in first turn OpenCode took 11.5k Setup: 1) Trimmed OpenCode (from usual 30k first turn to 11.5k) - 0 MCPs - 2 lightweight plugins (opencode-env-protect and openslimedit) - 8k char AGENTS.md - 11585 deepseek-v4-flash input tokens - for $0.0016 2) Vanilla Pi - 0 MCPs - 0 system prompt - 8k char AGENTS.md - 1114 kimi-k2.6 input tokens - for $0.0008 I think using better models with capped tokens per turn can keep usage nearly the same as uncontrolled DeepSeek V4 Flash? The challenge now is finding the sweet spot. Can't cap tokens if quality drops. We'll see. Video below - OpenCode vs Pi side by side "say hi back" first turns test, with OpenCode Go usage for each

raymel 👋

209,785 次观看 • 5 个月前

Let me give you a sneak peek at a tool I vibe-coded to simplify starting any new UI project—whether it’s a landing page or an app. All it needs as input is a hue and the amount of color in gray shades, and it instantly generates basic color tokens for both light and dark themes.

Let me give you a sneak peek at a tool I vibe-coded to simplify starting any new UI project—whether it’s a landing page or an app. All it needs as input is a hue and the amount of color in gray shades, and it instantly generates basic color tokens for both light and dark themes.

Roman Shamin

66,261 次观看 • 1 年前

Opus 5 is a BEAST. I tried Matt Shumer's viral prompt and it actually one shot a real FPS zombies style game. I tried something similar with Fable and Kimi K3 and it never got anywhere close to this. This cost me almost $400 in tokens and took about 3 hours.

Opus 5 is a BEAST. I tried Matt Shumer's viral prompt and it actually one shot a real FPS zombies style game. I tried something similar with Fable and Kimi K3 and it never got anywhere close to this. This cost me almost $400 in tokens and took about 3 hours.

Mark Gadala-Maria

18,782 次观看 • 3 个月前

This is GTB hospital in Delhi More than a thousand Students and Doctors took the streets to protest in solidarity with the Kolkata Victim . #JusticeForMoumita

This is GTB hospital in Delhi More than a thousand Students and Doctors took the streets to protest in solidarity with the Kolkata Victim . #JusticeForMoumita

Hemant Rajaura

68,757 次观看 • 2 年前

Most don't know (1) how easy it is to invert embedding vectors back into sentences, (2) this is a perfect task text diffusion models. Here's a 78M parameter model and live demo that recovers 80% of tokens from Qwen3-Embedding and EmbeddingGemma vectors. Works even on multilingual input.

Most don't know (1) how easy it is to invert embedding vectors back into sentences, (2) this is a perfect task text diffusion models. Here's a 78M parameter model and live demo that recovers 80% of tokens from Qwen3-Embedding and EmbeddingGemma vectors. Works even on multilingual input.

Jina AI

13,259 次观看 • 8 个月前

holy shit, i built a better Cluely. introducing: Obscura, a better, more performant Cluely. Obscura helps you during your meeting and interview calls, in real time, with more context and a built-in agent to websearch in less than a second. it is completely INVISIBLE to your screen recorder, screen share and cannot be detected during your google calls or zoom calls no matter what. some functionalities: 1. toggle visibility: so that only you can see it 2. panic kill: if you wanna kill it in a second 3. screenshot ocr: for screenshot capabilities 4. context window: pass any context in .md or .docx file for it to read before your calls and Obscura will understand it 5. performative built-in agent: to search for anything in less than a second. it uses Deepgram for STT model and Anthropic / OpenAI api keys to generate answers. it takes $1 for 20 hours of your calls as compared to cluely's $20/mo or $75/mo plan (but that depends on tokens consumed on both models). releasing a .dmg for people for people to use. access will be given to people who share this with their friends. I will roll out a referral program so that people can share it with others. feel free to give some feedback. thanks to rahat for being on call with me and testing it.

holy shit, i built a better Cluely. introducing: Obscura, a better, more performant Cluely. Obscura helps you during your meeting and interview calls, in real time, with more context and a built-in agent to websearch in less than a second. it is completely INVISIBLE to your screen recorder, screen share and cannot be detected during your google calls or zoom calls no matter what. some functionalities: 1. toggle visibility: so that only you can see it 2. panic kill: if you wanna kill it in a second 3. screenshot ocr: for screenshot capabilities 4. context window: pass any context in .md or .docx file for it to read before your calls and Obscura will understand it 5. performative built-in agent: to search for anything in less than a second. it uses Deepgram for STT model and Anthropic / OpenAI api keys to generate answers. it takes $1 for 20 hours of your calls as compared to cluely's $20/mo or $75/mo plan (but that depends on tokens consumed on both models). releasing a .dmg for people for people to use. access will be given to people who share this with their friends. I will roll out a referral program so that people can share it with others. feel free to give some feedback. thanks to rahat for being on call with me and testing it.

Naruto11.eth

15,760 次观看 • 7 个月前

Chrome DevTools now supports *individual* network request throttling! Folks have been asking for this for years! DevTools now allows developers to simulate slow network conditions for specific requests rather than the entire page. This helps in testing how a web application performs and handles issues when specific resources (like images, scripts, or API calls) are slow to load.

Chrome DevTools now supports individual network request throttling! Folks have been asking for this for years! DevTools now allows developers to simulate slow network conditions for specific requests rather than the entire page. This helps in testing how a web application performs and handles issues when specific resources (like images, scripts, or API calls) are slow to load.

Addy Osmani

281,833 次观看 • 9 个月前