Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

this is f*cking insane, Claude Code turned a technical drawing into a 3D model you can edit Dario Amodei (CEO of Anthropic) put 14 agents to work on a CAD application with 620,000 lines of code. the assignment was to make it build parts from drawings and check every... dimension it could read drop in a bracket drawing. get a model with editable holes, bends and mounting points change the hole spacing from 40 mm to 55 mm and the part rebuilds around the new measurement a model that looks right can still be impossible to manufacture one drawing left the wall thickness unspecified. the system flagged the missing measurement instead of quietly inventing it 34 of 40 sample parts passed dimension and solid-geometry checks. six needed clarification before they could be accepted give Claude Code an open-source CAD repository and a fully dimensioned sample drawing, then ask: "Build a drawing-to-model workflow inside this application 1. Extract dimensions, units and geometric constraints from the drawing. Link each measurement to its location and flag anything missing or ambiguous 2. Use the application's existing modeling API to construct an editable feature tree. Keep dimensions as named parameters rather than baking them into a mesh 3. Have a separate agent compare the model against the drawing. Check dimensions, hole positions and whether the result is a valid solid 4. Build a view showing the drawing beside the model. Let me change a parameter, rebuild the part and export it with a verification report" start with a simple part whose dimensions you can check by hand keep unresolved measurements visible. passing these checks does not establish material strength or fitness for a real assembly watch the drawing become an editable model belowshow more

Mika

1,381 subscribers

33,125 Aufrufe • vor 2 Tagen •via X (Twitter)

Bildung Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

A quick test of using 3d drawing with 6DOF controllers as an "instructor" for a generative AI process. There's so many powerful and fun new ways to create just around the corner.. Here I'm using Dreams, Krea and 3daistudio. The 3d model at the end of the video was generated from the Dreams+Krea output in just around 15 seconds. Only the model on the left is a "true" 3d model. #ai #madeindreams

A quick test of using 3d drawing with 6DOF controllers as an "instructor" for a generative AI process. There's so many powerful and fun new ways to create just around the corner.. Here I'm using Dreams, Krea and 3daistudio. The 3d model at the end of the video was generated from the Dreams+Krea output in just around 15 seconds. Only the model on the left is a "true" 3d model. #ai #madeindreams

Martin Nebelong

132,897 Aufrufe • vor 2 Jahren

Introducing Pages on Superset mobile Your agents can now deliver interactive demos, mockups, and reports as Pages you can review from your phone. Ask an agent to build a feature, compare designs, or investigate a problem. It can put the result in a Page with the context you need to review it and give feedback. A few ways to use it: • Feature work: try an interactive demo, then inspect the code in the PR. • Design reviews: compare mockups and before-and-after screenshots. • Investigations: explore findings, charts, and recommendations in one place. To try it, ask your agent to build a feature and create a Page for you to review.

Introducing Pages on Superset mobile Your agents can now deliver interactive demos, mockups, and reports as Pages you can review from your phone. Ask an agent to build a feature, compare designs, or investigate a problem. It can put the result in a Page with the context you need to review it and give feedback. A few ways to use it: • Feature work: try an interactive demo, then inspect the code in the PR. • Design reviews: compare mockups and before-and-after screenshots. • Investigations: explore findings, charts, and recommendations in one place. To try it, ask your agent to build a feature and create a Page for you to review.

Kiet

85,841 Aufrufe • vor 23 Tagen

Okay GPT-6 Astra might actually be the most insane thing we've ever seen for 3D. It drives Blender by itself. It clicks the same buttons a real artist would. All it needs is a drawing or an image to start. From there it gives you thousands of fully editable detailed objects with geometry so clean it hurts. Anyone can build an entire 3D world just by talking now. What is happening.

Okay GPT-6 Astra might actually be the most insane thing we've ever seen for 3D. It drives Blender by itself. It clicks the same buttons a real artist would. All it needs is a drawing or an image to start. From there it gives you thousands of fully editable detailed objects with geometry so clean it hurts. Anyone can build an entire 3D world just by talking now. What is happening.

ImagineArt

21,526 Aufrufe • vor 1 Monat

Another big drop from Hunyuan3D by Tencent. Hunyuan3D-Buffalo 1.0 is not just another text-to-3D model. The more interesting part is its approach to 3D understanding and editing. It can understand separate parts of a complete asset, extract them, remove or replace them with prompts, and edit only the selected region while keeping the rest of the geometry unchanged. That feels like a much more useful direction for 3D AI in general: not just generating a mesh once, but actually understanding its structure and letting you continue working with it. The model combines Qwen-VL, TRELLIS and Hunyuan3D, and was trained on an impressive dataset of 87 million 3D samples. Source:

Another big drop from Hunyuan3D by Tencent. Hunyuan3D-Buffalo 1.0 is not just another text-to-3D model. The more interesting part is its approach to 3D understanding and editing. It can understand separate parts of a complete asset, extract them, remove or replace them with prompts, and edit only the selected region while keeping the rest of the geometry unchanged. That feels like a much more useful direction for 3D AI in general: not just generating a mesh once, but actually understanding its structure and letting you continue working with it. The model combines Qwen-VL, TRELLIS and Hunyuan3D, and was trained on an impressive dataset of 87 million 3D samples. Source:

Stefan 3D AI

36,409 Aufrufe • vor 2 Monaten

One idea, a playable game. Nex-N2.5 Pro drove the whole thing rough idea →2D sketch →Blender 3D scene →finished game. It wasn't prompting from a distance it opened the tools, watched what changed on screen, and kept going the demos keep landing on the same question: can it actually use a computer? not generate an image of a kitchen call Blender MCP, build the kitchen, render it, read the result. Not describe a CAD part script it, adjust parameters, hit an export wall, switch to the GUI, hand off the drawing try it free on openRouter for a limited time: #NexN2 #Blender #ComputerUse #OpenRouter

One idea, a playable game. Nex-N2.5 Pro drove the whole thing rough idea →2D sketch →Blender 3D scene →finished game. It wasn't prompting from a distance it opened the tools, watched what changed on screen, and kept going the demos keep landing on the same question: can it actually use a computer? not generate an image of a kitchen call Blender MCP, build the kitchen, render it, read the result. Not describe a CAD part script it, adjust parameters, hit an export wall, switch to the GUI, hand off the drawing try it free on openRouter for a limited time: #NexN2 #Blender #ComputerUse #OpenRouter

Nahid

90,852 Aufrufe • vor 1 Monat

HydraFusion Explained. Part I: How does the Copilot engine know what to optimize for? Your prompt is evaluated across 4 dimensions: ➡ Does it require deep reasoning? (aka. reasoning depth) ➡ Is it a sophisticated problem? (aka. code generation complexity) ➡ Is it untangling a complicated mess? (aka. debugging difficulty) ➡ Is it dominated by tool-use? (aka. tool orchestration needs) Based on this evaluation, a HyDRA score is assigned to determine the capability profile your task needs the most and to establish a quality bar. Part II: How does it choose a model? Note: It doesn' t pick one model to handle the entire job e2e, (that's Auto mode). Instead, it selects 1 of 3 execution workflows and assigns the best model at different stages based on the HyDRA score: 1️⃣ Single ⚙️ How it works: A single model completes the task from start to finish. ⚖️ Rationale: The task comfortably meets the quality bar with one model. Multi-model orchestration would add latency and cost with no meaningful quality gain. 2️⃣ Cascade ⚙️ How it works: A lightweight, cost-efficient model generates the solution. This draft is evaluated against a quality gate and if it falls short of the quality bar, the entire task escalates to a stronger, frontier model. ⚖️ Rationale: Only bring in the big guns when there is concrete evidence that a lightweight model won't meet the quality threshold. 3️⃣ Critique ⚙️ How it works: A lightweight model drafts the initial code and tool interactions. An independent, read-only frontier model reviews that draft and provides feedback. The original lightweight model then performs any targeted revision(s) before the final response is sent to the user. ⚖️ Rationale: Writing code (output tokens) is expensive while reviewing code (input tokens) is cheap. Instead of incurring the cost of a powerhouse writing hundreds of lines from scratch, a cost-efficient model writes the first draft, and the frontier model just reviews it and points out fixes. HydraFusion is available in experimental preview on the GitHub Copilot CLI: /experimental on, /model and select Hydrafusion (Research Preview)

Julia Muiruri

13,000 Aufrufe • vor 1 Monat

this is f*cking gold Google engineers explained how to make AI rewrite an agent’s system prompt against failing tests. a failed check becomes the next repair job. the regression suite watches for whatever that repair breaks. even the instructions become something you can test and improve. Agent = Model + Harness. the independently compiled page here maps the broader system into six parts: guides carry project rules, constraints and lessons from past failures > sensors check the work through tests, linters and validators > the loop runs the task, checks the result, retries within limits and escalates > memory preserves state, artifacts and decisions across runs > permissions control tools, writes and actions requiring approval > observability records what happened, what it cost and where it failed an agent changes a build file and announces “done.” the check looks for evidence that it actually ran the validator. if that behavior is missing, you have a specific failure to target. adjust the instructions. rerun the evaluations. check whether the improvement holds without breaking existing behavior. now a prompt change has a test history. a recurring mistake has a regression check. the next model upgrade has something concrete to pass. bookmark the diagram. give “done” a test it has to pass.

this is f*cking gold Google engineers explained how to make AI rewrite an agent’s system prompt against failing tests. a failed check becomes the next repair job. the regression suite watches for whatever that repair breaks. even the instructions become something you can test and improve. Agent = Model + Harness. the independently compiled page here maps the broader system into six parts: guides carry project rules, constraints and lessons from past failures > sensors check the work through tests, linters and validators > the loop runs the task, checks the result, retries within limits and escalates > memory preserves state, artifacts and decisions across runs > permissions control tools, writes and actions requiring approval > observability records what happened, what it cost and where it failed an agent changes a build file and announces “done.” the check looks for evidence that it actually ran the validator. if that behavior is missing, you have a specific failure to target. adjust the instructions. rerun the evaluations. check whether the improvement holds without breaking existing behavior. now a prompt change has a test history. a recurring mistake has a regression check. the next model upgrade has something concrete to pass. bookmark the diagram. give “done” a test it has to pass.

NO1ennn

72,042 Aufrufe • vor 1 Monat

Blender just became a prompt box. With Kimi K3 connected through Blender MCP, you can describe a scene like this in plain English and let the model handle the ugly part: terrain, buildings, lighting, materials, camera movement, animation, and the Python scripts holding everything together. The interesting part is not the first render. Kimi can inspect the result, notice that the camera clips through a tree or the city looks suspiciously like plastic, then edit the actual Blender scene and render it again instead of restarting from zero. You still need taste, because “make it cinematic” remains one of humanity’s least useful instructions. But the distance between an empty Blender file and a fully editable 3D world just got embarrassingly small. A few years ago, creating this meant weeks of modeling, scripting, lighting, and animation. Now you can describe the world, watch Kimi build it, and spend your time fixing the final 10% instead of manually constructing the first 90%. Kimi K3 + Blender MCP is basically text-to-3D without trapping the result inside a useless generated video. Every object, material, light, camera, and keyframe stays editable.

Blender just became a prompt box. With Kimi K3 connected through Blender MCP, you can describe a scene like this in plain English and let the model handle the ugly part: terrain, buildings, lighting, materials, camera movement, animation, and the Python scripts holding everything together. The interesting part is not the first render. Kimi can inspect the result, notice that the camera clips through a tree or the city looks suspiciously like plastic, then edit the actual Blender scene and render it again instead of restarting from zero. You still need taste, because “make it cinematic” remains one of humanity’s least useful instructions. But the distance between an empty Blender file and a fully editable 3D world just got embarrassingly small. A few years ago, creating this meant weeks of modeling, scripting, lighting, and animation. Now you can describe the world, watch Kimi build it, and spend your time fixing the final 10% instead of manually constructing the first 90%. Kimi K3 + Blender MCP is basically text-to-3D without trapping the result inside a useless generated video. Every object, material, light, camera, and keyframe stays editable.

Rina

264,903 Aufrufe • vor 2 Monaten

i gave bonsai, 3.9 gb model one spec to build a working rotary engine autonomously, using hermes agent. all on rtx 3060ti 8gb. watch it go from a black canvas to a running wankel. the epitrochoid housing, a triangular rotor turning at exactly a third of shaft speed, three chambers firing intake, compression, power, exhaust, and ten geometry tests it had to pass to count as done. a provably correct simulation, not a drawing. then i left it alone and it worked. wrote the code, ran its own tests, hit failures, read them, fixed its own lines, until the engine spun and all ten went green. a real agent loop, and i watched every step. it broke three times getting here and clawed back every one. that honest part is in the reply. and this is a sneak. i've been building a show around runs exactly like this, and it's almost ready. more very soon. just getting started.

i gave bonsai, 3.9 gb model one spec to build a working rotary engine autonomously, using hermes agent. all on rtx 3060ti 8gb. watch it go from a black canvas to a running wankel. the epitrochoid housing, a triangular rotor turning at exactly a third of shaft speed, three chambers firing intake, compression, power, exhaust, and ten geometry tests it had to pass to count as done. a provably correct simulation, not a drawing. then i left it alone and it worked. wrote the code, ran its own tests, hit failures, read them, fixed its own lines, until the engine spun and all ten went green. a real agent loop, and i watched every step. it broke three times getting here and clawed back every one. that honest part is in the reply. and this is a sneak. i've been building a show around runs exactly like this, and it's almost ready. more very soon. just getting started.

Sudo su

35,069 Aufrufe • vor 2 Monaten

claude opus 5.5 completely solved three.js it took a floor plan and turned it into a 3d house you can walk through home builders pay $5k to $10k per floor plan for this, so buyers can tour a home before it's built and change things if needed here's how to automate client acquisition: > scrape builder sites for plans with no model home > give the plan and their photos to opus 5.5 > it builds the house to scale as a walkable site > send the builder their own plan before you pitch > charge per plan to do the rest of their catalog > rerun it on every new plan they launch give this prompt to your agent to get started: "here's a floor plan and photos for one house. build it as a three.js site i can walk through in the browser, to scale from the plan, with every room, finish and piece of furniture matched to the photos. the bar is the photos themselves. split the house into pieces you can judge separately, and give each one a builder and a separate critic with fresh context. the critic compares a screenshot of our room with the matching photo, picks the better one, names the biggest gap, and sends it back. keep looping until ours wins or i stop you, and tell me anything you couldn't match."

claude opus 5.5 completely solved three.js it took a floor plan and turned it into a 3d house you can walk through home builders pay $5k to $10k per floor plan for this, so buyers can tour a home before it's built and change things if needed here's how to automate client acquisition: > scrape builder sites for plans with no model home > give the plan and their photos to opus 5.5 > it builds the house to scale as a walkable site > send the builder their own plan before you pitch > charge per plan to do the rest of their catalog > rerun it on every new plan they launch give this prompt to your agent to get started: "here's a floor plan and photos for one house. build it as a three.js site i can walk through in the browser, to scale from the plan, with every room, finish and piece of furniture matched to the photos. the bar is the photos themselves. split the house into pieces you can judge separately, and give each one a builder and a separate critic with fresh context. the critic compares a screenshot of our room with the matching photo, picks the better one, names the biggest gap, and sends it back. keep looping until ours wins or i stop you, and tell me anything you couldn't match."

Chris

305,958 Aufrufe • vor 4 Tagen

This is f*cking insane. This tip saved me thousands of dollars. run Opus 5.5, Sonnet 5.5, and Fable 5.1 together, and stop burning Opus on work it was never needed for. the whole idea in one line: the strong model plans, the mid-tier model executes, Fable stays quiet until it's actually needed. roles, broken down: Opus 5.5, high effort, owns the plan and ships the final code Sonnet 5.5, medium effort, splits into explorer (reads the codebase), worker (edits files, runs tests), researcher (pulls docs) Fable 5.1, called through /advisor fable, reads everything happening in the session but stays silent unless something's actually wrong three moments where Fable speaks: → a plan goes out: is this actually the right call? → the same failure shows up again: is the search going nowhere? → the task gets marked finished: did something get skipped? Jev engineering does the same thing one level down. the forks that don't need real thought, which file, which tool, keep going or stop, go straight to Jev and come back in under half a second. the big models only ever see the forks that genuinely need a decision. anyone still running one model for everything is paying Opus prices to decide whether a file exists. drop this into Claude Code: "Rebuild my Claude Code setup around this structure: Look through ~/.claude/agents and .claude/agents for subagents already covering explorer, worker, and researcher. Only create new ones for roles that are missing. Set model: sonnet, effort: medium on each. If an existing subagent is locked to a different model, leave it as is and just list it. In ~/.claude/settings.json, set effortLevel to high and advisorModel to fable. Check for anything disabling the advisor, CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, or anything blocking feature-flag fetches, plus CLAUDE_CODE_EFFORT_LEVEL, which can override subagent effort settings. Report what you find. Don't change any of it yet. Add one line to ~/.claude/CLAUDE.md: check in with the advisor before a big plan, when the same error shows up twice, and before marking a long task done. Show every change as a diff first. Wait for my go-ahead before touching anything."

This is f*cking insane. This tip saved me thousands of dollars. run Opus 5.5, Sonnet 5.5, and Fable 5.1 together, and stop burning Opus on work it was never needed for. the whole idea in one line: the strong model plans, the mid-tier model executes, Fable stays quiet until it's actually needed. roles, broken down: Opus 5.5, high effort, owns the plan and ships the final code Sonnet 5.5, medium effort, splits into explorer (reads the codebase), worker (edits files, runs tests), researcher (pulls docs) Fable 5.1, called through /advisor fable, reads everything happening in the session but stays silent unless something's actually wrong three moments where Fable speaks: → a plan goes out: is this actually the right call? → the same failure shows up again: is the search going nowhere? → the task gets marked finished: did something get skipped? Jev engineering does the same thing one level down. the forks that don't need real thought, which file, which tool, keep going or stop, go straight to Jev and come back in under half a second. the big models only ever see the forks that genuinely need a decision. anyone still running one model for everything is paying Opus prices to decide whether a file exists. drop this into Claude Code: "Rebuild my Claude Code setup around this structure: Look through ~/.claude/agents and .claude/agents for subagents already covering explorer, worker, and researcher. Only create new ones for roles that are missing. Set model: sonnet, effort: medium on each. If an existing subagent is locked to a different model, leave it as is and just list it. In ~/.claude/settings.json, set effortLevel to high and advisorModel to fable. Check for anything disabling the advisor, CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, or anything blocking feature-flag fetches, plus CLAUDE_CODE_EFFORT_LEVEL, which can override subagent effort settings. Report what you find. Don't change any of it yet. Add one line to ~/.claude/CLAUDE.md: check in with the advisor before a big plan, when the same error shows up twice, and before marking a long task done. Show every change as a diff first. Wait for my go-ahead before touching anything."

rvaniaaa

253,200 Aufrufe • vor 15 Tagen

New open-source agent harness just landed! I got early access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.

New open-source agent harness just landed! I got early access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.

elvis

11,303 Aufrufe • vor 1 Monat

I went a little overboard with Codex last week and burned through my entire weekly allowance in two days. Luckily, my quota reset today. Otherwise, I’m not sure what I would’ve done. It got me thinking: instead of asking one large model to handle everything from start to finish, why not let a stronger model plan the project and review the work, while a model built for execution handles the day-to-day implementation? So I tried it. The result was better than I expected. I used GPT-5.6 Sol in Codex as the decision-maker, then ran Ling-3.0-flash from Ant Ling inside OpenCode as the execution engine. Together, they built a small 3D farming game. Before writing any code, I had Codex create four documents: SPEC.md defined the product scope and the lines we couldn’t cross. ARCHITECTURE.md laid out the isometric coordinate system, state machine, and module boundaries. TASKS.md broke the project into small jobs Ling could tackle one at a time. ACCEPTANCE.md explained how each step would be tested and what “done” actually meant. Then I gave Ling a very straightforward role: You are the execution model for this project. Read all four documents before you begin. Work only on the task assigned for this round. When you’re done, run typecheck, test, and build. If anything fails, read the error, fix it, and run the checks again. Do not move on to the next task early. Ling handled dependency installation, project structure, strict TypeScript configuration, test setup, and a production build in 6 minutes and 3 seconds. It ran into issues with the Vite test config, a TS6310 error, and a missing jsdom dependency along the way. Instead of stopping at the first error, it kept reading the logs and fixing the problems until all three checks passed. The speed was honestly hard to believe. If you exclude the time spent waiting on tools, it was producing more than 100 tokens per second. That made the whole development loop feel noticeably faster. After this experiment, I’m planning to keep using the same workflow. If the task is small, there’s no reason to call an expensive planning model for every single step. If the task is large, handing the entire project to a Flash model in one prompt isn’t a great idea either. The setup that makes more sense to me is: Use a more capable model such as Codex to explore the project, make architectural decisions, and break the work down. Put the constraints into specs, schemas, types, and tests instead of leaving them buried in chat history. Give Ling-3.0-flash a steady stream of clear, verifiable implementation tasks. Report bugs with structured context and actual error logs, rather than saying, “It still doesn’t work.” Bring Codex back in for architecture reviews, visual checks, and changes that affect multiple parts of the project. The point of this setup isn’t to give AI a big “build the whole project” button. It’s to turn software development into a pipeline with a much more sensible cost structure: Codex figures out the plan, sets the boundaries, and catches problems. Ling-3.0-flash moves quickly, calls tools reliably, and works through well-defined tasks at scale. For agent workflows that involve lots of repetitive edits, production tasks, and tool calls, this may be a more practical answer than simply using the biggest model for everything.

I went a little overboard with Codex last week and burned through my entire weekly allowance in two days. Luckily, my quota reset today. Otherwise, I’m not sure what I would’ve done. It got me thinking: instead of asking one large model to handle everything from start to finish, why not let a stronger model plan the project and review the work, while a model built for execution handles the day-to-day implementation? So I tried it. The result was better than I expected. I used GPT-5.6 Sol in Codex as the decision-maker, then ran Ling-3.0-flash from Ant Ling inside OpenCode as the execution engine. Together, they built a small 3D farming game. Before writing any code, I had Codex create four documents: SPEC.md defined the product scope and the lines we couldn’t cross. ARCHITECTURE.md laid out the isometric coordinate system, state machine, and module boundaries. TASKS.md broke the project into small jobs Ling could tackle one at a time. ACCEPTANCE.md explained how each step would be tested and what “done” actually meant. Then I gave Ling a very straightforward role: You are the execution model for this project. Read all four documents before you begin. Work only on the task assigned for this round. When you’re done, run typecheck, test, and build. If anything fails, read the error, fix it, and run the checks again. Do not move on to the next task early. Ling handled dependency installation, project structure, strict TypeScript configuration, test setup, and a production build in 6 minutes and 3 seconds. It ran into issues with the Vite test config, a TS6310 error, and a missing jsdom dependency along the way. Instead of stopping at the first error, it kept reading the logs and fixing the problems until all three checks passed. The speed was honestly hard to believe. If you exclude the time spent waiting on tools, it was producing more than 100 tokens per second. That made the whole development loop feel noticeably faster. After this experiment, I’m planning to keep using the same workflow. If the task is small, there’s no reason to call an expensive planning model for every single step. If the task is large, handing the entire project to a Flash model in one prompt isn’t a great idea either. The setup that makes more sense to me is: Use a more capable model such as Codex to explore the project, make architectural decisions, and break the work down. Put the constraints into specs, schemas, types, and tests instead of leaving them buried in chat history. Give Ling-3.0-flash a steady stream of clear, verifiable implementation tasks. Report bugs with structured context and actual error logs, rather than saying, “It still doesn’t work.” Bring Codex back in for architecture reviews, visual checks, and changes that affect multiple parts of the project. The point of this setup isn’t to give AI a big “build the whole project” button. It’s to turn software development into a pipeline with a much more sensible cost structure: Codex figures out the plan, sets the boundaries, and catches problems. Ling-3.0-flash moves quickly, calls tools reliably, and works through well-defined tasks at scale. For agent workflows that involve lots of repetitive edits, production tasks, and tool calls, this may be a more practical answer than simply using the biggest model for everything.

雪踏乌云

23,107 Aufrufe • vor 2 Monaten

50% cheaper Claude inference with just one line of code change! - Remove → model="claude-opus-4-8" - Add → model="ship-like/claude-opus-4-8" I verified the cost saving in my own terminal by invoking the same Anthropic model with the same prompt. The underlying engineering by Ship is actually interesting, and the patterns can be used in any production LLM stack. Essentially, a trained model is a frozen artifact. Every request performs the same forward-pass, whether it extracts a date or refactors a module, because the compute decision was made at training time, before the request existed. Ship makes that decision at inference time instead. After seeing a request, it searches over executions, involving single models, cascades, ensembles, or harnesses with tools, and serves the cheapest one that will match the reference model's quality. This is not a basic router, because picking a cheaper model per query doesn't ensure the cheaper model preserves the original's behavior, like output shape, tool-call patterns, and refusals. Ship measures this equivalence directly. Outputs stay distributionally indistinguishable from the reference model, not token-identical, since two calls to the same model already differ, but they are indistinguishable in capability and behavior. Of course, some requests execute cheaply and some cost Ship more than the customer pays, but the price per request is still a flat 50% off either way, so the execution-cost variance moves off the application's bill entirely. The video below depicts the cost savings and output in my real invocation, and I partnered with the team to put this together.

50% cheaper Claude inference with just one line of code change! - Remove → model="claude-opus-4-8" - Add → model="ship-like/claude-opus-4-8" I verified the cost saving in my own terminal by invoking the same Anthropic model with the same prompt. The underlying engineering by Ship is actually interesting, and the patterns can be used in any production LLM stack. Essentially, a trained model is a frozen artifact. Every request performs the same forward-pass, whether it extracts a date or refactors a module, because the compute decision was made at training time, before the request existed. Ship makes that decision at inference time instead. After seeing a request, it searches over executions, involving single models, cascades, ensembles, or harnesses with tools, and serves the cheapest one that will match the reference model's quality. This is not a basic router, because picking a cheaper model per query doesn't ensure the cheaper model preserves the original's behavior, like output shape, tool-call patterns, and refusals. Ship measures this equivalence directly. Outputs stay distributionally indistinguishable from the reference model, not token-identical, since two calls to the same model already differ, but they are indistinguishable in capability and behavior. Of course, some requests execute cheaply and some cost Ship more than the customer pays, but the price per request is still a flat 50% off either way, so the execution-cost variance moves off the application's bill entirely. The video below depicts the cost savings and output in my real invocation, and I partnered with the team to put this together.

Akshay 🚀

64,482 Aufrufe • vor 2 Monaten

We are getting absurdly close to the point where “learning Blender” means learning how to direct an AI. A shot like this looks soft and playful on the surface, but under it is the usual 3D pain: modeling, layout, materials, lighting, atmosphere, animation, and endless tiny fixes until the frame stops looking dead. That is why Kimi K3 matters. With Blender MCP, you can describe a scene like a robotic goat walking through a dreamy field and let the model help build the environment, place the camera, shape the materials, script the motion, and iterate inside the real Blender project. The real shift is not text-to-image. It is text-to-workflow. Kimi K3 does not just give you a pretty output and disappear. It can help move the actual scene from rough setup to something that looks art-directed. Soon the hardest part of 3D will not be the software. It will be whether your imagination is good enough to deserve tools like this.

We are getting absurdly close to the point where “learning Blender” means learning how to direct an AI. A shot like this looks soft and playful on the surface, but under it is the usual 3D pain: modeling, layout, materials, lighting, atmosphere, animation, and endless tiny fixes until the frame stops looking dead. That is why Kimi K3 matters. With Blender MCP, you can describe a scene like a robotic goat walking through a dreamy field and let the model help build the environment, place the camera, shape the materials, script the motion, and iterate inside the real Blender project. The real shift is not text-to-image. It is text-to-workflow. Kimi K3 does not just give you a pretty output and disappear. It can help move the actual scene from rough setup to something that looks art-directed. Soon the hardest part of 3D will not be the software. It will be whether your imagination is good enough to deserve tools like this.

Rina

45,651 Aufrufe • vor 2 Monaten

Stop manually making flowcharts. Try this GitHub skill with Claude Code. I found an open-source tool that turns a plain description of a software system into a clean, interactive diagram. It's called Archify, and it has more than 66,000 stars on GitHub. You add it as a skill to an AI coding agent like Claude Code, Codex, Cursor, or OpenCode. Then you describe the system in the chat in one line. Archify turns that line into a single HTML file you can open, click through, and send to anyone. If you point your agent at an existing codebase, it reads the code and Archify maps how the pieces connect. I liked five things about it. 1. It makes five kinds of diagrams, including system maps and step-by-step workflows. 2. You can click any piece and trace what connects to it. 3. It checks the layout before it hands the diagram over, so arrows don't pile up and labels don't overlap. 4. It has dark and light themes and exports to PNG, SVG, or a short video. 5. You keep editing in plain chat, with requests like "add Redis" or "highlight the rollback path." You don't need a codebase to use it. A description in the chat is enough.

Stop manually making flowcharts. Try this GitHub skill with Claude Code. I found an open-source tool that turns a plain description of a software system into a clean, interactive diagram. It's called Archify, and it has more than 66,000 stars on GitHub. You add it as a skill to an AI coding agent like Claude Code, Codex, Cursor, or OpenCode. Then you describe the system in the chat in one line. Archify turns that line into a single HTML file you can open, click through, and send to anyone. If you point your agent at an existing codebase, it reads the code and Archify maps how the pieces connect. I liked five things about it. 1. It makes five kinds of diagrams, including system maps and step-by-step workflows. 2. You can click any piece and trace what connects to it. 3. It checks the layout before it hands the diagram over, so arrows don't pile up and labels don't overlap. 4. It has dark and light themes and exports to PNG, SVG, or a short video. 5. You keep editing in plain chat, with requests like "add Redis" or "highlight the rollback path." You don't need a codebase to use it. A description in the chat is enough.

Alex Veremeyenko

61,928 Aufrufe • vor 28 Tagen

Claude Code tip: if Opus 5.5 is already your main model, Fable 5.1 has been sitting idle this whole time. wire it in with /advisor start it with /advisor fable Opus 5.5 writes every line. Fable 5.1 reads the whole session, every tool call, and says nothing until one of three moments: → a plan gets proposed: is this actually the right move, or just the first one? → the same error comes back twice: is the search stuck, or is this a dead end? → the task gets marked done: what got missed while it was moving fast? Opus 5.5 ships. Fable 5.1 catches what would've shipped broken. Jev engineering makes the same move one layer down: forks that don't need a real thinker, which file, which tool, retry or give up, get routed to Jev and answered in under half a second. the expensive model only ever sees the forks that genuinely split. the tree this runs on: > Opus 5.5, high effort, owns the main session > explorer, medium effort, reads the code > worker, medium effort, edits and runs tests > researcher, medium effort, pulls the docs > Fable 5.1 outside all of it, on call, never writing a line itself drop the tree and this prompt into Claude Code: "Rebuild my Claude Code setup around this tree: 1. Look in ~/.claude/agents and .claude/agents for subagents that already cover explorer, worker and researcher. Draft new ones only for roles that are missing. Set each to model: opus, effort: medium. If an existing subagent is pinned to a different model, list it, don't touch it. 2. Set the main session's effortLevel to high in ~/.claude/settings.json, and set advisorModel to fable. 3. Check for anything disabling the advisor: CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, any variable blocking feature-flag fetches, and CLAUDE_CODE_EFFORT_LEVEL, which overrides subagent effort. Report what you find. Change nothing yet. 4. Add one line to ~/.claude/CLAUDE.md: consult the advisor before a large plan, when an error repeats, and before marking a long task done. Show every change as a diff first. Don't touch anything until I say go."

Claude Code tip: if Opus 5.5 is already your main model, Fable 5.1 has been sitting idle this whole time. wire it in with /advisor start it with /advisor fable Opus 5.5 writes every line. Fable 5.1 reads the whole session, every tool call, and says nothing until one of three moments: → a plan gets proposed: is this actually the right move, or just the first one? → the same error comes back twice: is the search stuck, or is this a dead end? → the task gets marked done: what got missed while it was moving fast? Opus 5.5 ships. Fable 5.1 catches what would've shipped broken. Jev engineering makes the same move one layer down: forks that don't need a real thinker, which file, which tool, retry or give up, get routed to Jev and answered in under half a second. the expensive model only ever sees the forks that genuinely split. the tree this runs on: > Opus 5.5, high effort, owns the main session > explorer, medium effort, reads the code > worker, medium effort, edits and runs tests > researcher, medium effort, pulls the docs > Fable 5.1 outside all of it, on call, never writing a line itself drop the tree and this prompt into Claude Code: "Rebuild my Claude Code setup around this tree: 1. Look in ~/.claude/agents and .claude/agents for subagents that already cover explorer, worker and researcher. Draft new ones only for roles that are missing. Set each to model: opus, effort: medium. If an existing subagent is pinned to a different model, list it, don't touch it. 2. Set the main session's effortLevel to high in ~/.claude/settings.json, and set advisorModel to fable. 3. Check for anything disabling the advisor: CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, any variable blocking feature-flag fetches, and CLAUDE_CODE_EFFORT_LEVEL, which overrides subagent effort. Report what you find. Change nothing yet. 4. Add one line to ~/.claude/CLAUDE.md: consult the advisor before a large plan, when an error repeats, and before marking a long task done. Show every change as a diff first. Don't touch anything until I say go."

Ryven

36,121 Aufrufe • vor 16 Tagen

Gemini 4 argon drawing Frieren in SVG asked model to make me an svg of frieren with a reference and give me final output with bg music too > took 1 hour 27 mins > used ~26.35M tokens > cost $5.19 in devin and here's the output it made, you guys decide if it cooked or nah

Gemini 4 argon drawing Frieren in SVG asked model to make me an svg of frieren with a reference and give me final output with bg music too > took 1 hour 27 mins > used ~26.35M tokens > cost $5.19 in devin and here's the output it made, you guys decide if it cooked or nah

J A Z I I

19,290 Aufrufe • vor 4 Tagen

If you are trying to understand where AI agents are going, learn harness engineering. A capable model is only one part of an agent system. Once the model begins reading files, calling tools, modifying state and working across many steps, the quality of the system depends increasingly on the software around it. Consider a coding agent working through a large repository. The model can decide that it needs to inspect a file, search for a symbol, make an edit or run a test, but those decisions do not execute themselves. The surrounding runtime has to decide which resources are available, whether the requested action is permitted, how the operation should be performed, what result should be retained, and what information should be presented to the model on the next step. This becomes harder as the run gets longer. As history accumulates, replaying everything can become costly and less effective. The harness has to decide what should remain in context, what should be summarized or retrieved later, and what belongs in persistent state outside the context window. Execution has similar problems. A long-running agent may need to survive an interruption, avoid repeating completed work, enforce permissions around consequential actions, and preserve enough history to reconstruct what happened when the final result is wrong. These are harness problems. The harness is the layer that manages context, tools, execution, state, checkpoints, limits and traces around the model. Harness engineering is the work of designing and improving that layer. Engineers inspect execution traces, evaluate agents on representative tasks, look for recurring failure modes, and then change things such as context selection, tool interfaces, state handling or execution controls. That last part matters because agent failures are often not fixed by changing the model. Sometimes the useful change is in what the model sees, how a tool is exposed, what state is preserved, or what the runtime does after a failed step. As agents take on longer tasks, the demands on this surrounding software grow. Model capability remains essential, but harness engineering is what turns that capability into an execution process that can be controlled, inspected, tested and improved.

If you are trying to understand where AI agents are going, learn harness engineering. A capable model is only one part of an agent system. Once the model begins reading files, calling tools, modifying state and working across many steps, the quality of the system depends increasingly on the software around it. Consider a coding agent working through a large repository. The model can decide that it needs to inspect a file, search for a symbol, make an edit or run a test, but those decisions do not execute themselves. The surrounding runtime has to decide which resources are available, whether the requested action is permitted, how the operation should be performed, what result should be retained, and what information should be presented to the model on the next step. This becomes harder as the run gets longer. As history accumulates, replaying everything can become costly and less effective. The harness has to decide what should remain in context, what should be summarized or retrieved later, and what belongs in persistent state outside the context window. Execution has similar problems. A long-running agent may need to survive an interruption, avoid repeating completed work, enforce permissions around consequential actions, and preserve enough history to reconstruct what happened when the final result is wrong. These are harness problems. The harness is the layer that manages context, tools, execution, state, checkpoints, limits and traces around the model. Harness engineering is the work of designing and improving that layer. Engineers inspect execution traces, evaluate agents on representative tasks, look for recurring failure modes, and then change things such as context selection, tool interfaces, state handling or execution controls. That last part matters because agent failures are often not fixed by changing the model. Sometimes the useful change is in what the model sees, how a tool is exposed, what state is preserved, or what the runtime does after a failed step. As agents take on longer tasks, the demands on this surrounding software grow. Model capability remains essential, but harness engineering is what turns that capability into an execution process that can be controlled, inspected, tested and improved.

Tech with Mak

49,737 Aufrufe • vor 21 Tagen