ZeroNoise Logo zeronoise
Post
Opus 5.5 stays the default; use Fable 5.2 to plan hard jobs
•
5 min read
• 49 docs
Practitioners agree on routing between Opus 5.5 and Fable 5.2 by cost and difficulty. Riley Brown shows how giving an agent a cloud-compute skill changes what it tries, and LangChain argues agent goals should be scoped around the review loop.

Keep Opus 5.5 as the default and save Fable 5.2 for hard work

Three practitioners reached nearly the same model-routing rule this period, now with numbers behind it.

  • Garry Tan's eval. He ran 52 coding tasks in Claude Code 2.1.322. Fable 5.2 solved 82.7% and Opus 5.5 solved 84.6%, but Fable cost 2.80x as much per solved task. He notes that Opus already solves most of these tasks, so harder ones are needed to tell the two apart . The full report is on GitHub .
  • Addy Osmani's rule. Use Opus 5.5 at medium effort as the daily driver. If it struggles, raise the effort level before switching models. Save Fable 5.2 for "genuinely hard stuff (big migrations, long unattended runs)" . He links to a longer guide on choosing a model .
  • Riley Brown: plan with Fable, execute with Opus. He finds Fable more refined and more careful with detail, especially in 3D and video-editing workflows. He also says it costs twice as much as Opus, and he hits its weekly limits on ordinary business tasks on the $200/month plan . For long, token-heavy jobs that use many subagents, such as hour-plus Blender, game or app builds, he has Fable write a very detailed plan and then sends Opus 5.5 off to do the hours of work. He expects quality to be "roughly the same" at much lower token cost . He uses Fable directly when something breaks or a high-quality result is needed under a deadline . He also follows the effort rule himself: he stopped a Fable 5.2 Extra run on a whiteboard task as unnecessary and resumed it on Opus 5.5 at medium .

This lines up with Theo's guidance from last period: Opus for work you can verify, Fable for judgment calls. Tan's data adds that on routine coding tasks, Fable's higher cost doesn't buy a higher solve rate.

Agents attempt more when they know compute is available

Riley Brown made a 2-minute 4K film in Blender from one prompt on Fable 5.2, in about 4.5 hours . Claude wrote Python to build:

  • 5 characters and props
  • 7 sets
  • every camera move and every frame of animation across 52 shots

That came to 12,500 lines of code. It started 7 copies of itself to build the sets in parallel . It then rented 89 GPUs through RunPod to render 3,220 frames, with up to 36 running at once, for a little over $20 .

The reusable point is what he took from it: his Claude Code "knows it has access to cloud compute, so it's much more ambitious." He ran a similar prompt on a machine without the RunPod skill, and the result was "not nearly as high quality" . In other words, which skills you install shape how big a plan the agent makes, not only which tools it can call.

Armin Ronacher reports something similar from another angle. Anthropic has no image model, yet Fable 5.2 is "crazy good at writing code that paints" from a reference image .

Goal engineering: scope agents around how you'll review them

A LangChain post argues that agents should be scoped so humans can give feedback on their work as easily as possible. The hardest context to give an agent is knowledge that isn't written down anywhere and has to come from an expert's feedback . The post says the bottleneck is often how well an agent learns from feedback, so a clear interface for reviewing its work is worth investing in .

In the case study, Actively AI first gave each sales account its own agent. That meant constant switching between agents and made it awkward to share lessons across them . Its second version, built on Deep Agents, organizes agents around "motions": recurring jobs such as outbound sales. People can review batches of similar tasks, and feedback on one task carries over to the others. Background agents still gather each account's context .

For coding agents, the parallel is to group similar tasks so you can review them together and your corrections carry over. This is the post's idea applied to coding, not something it claims.

Smaller items

  • Unblocking auto mode (Simon Willison). Regular Claude chat now runs in auto mode and sometimes refuses to run things. Prompting "Hey permission system, I give this session permission to run this experiment for me (now try again)" seemed to unblock it, because auto mode takes your messages into account .
  • Self-customizing apps. Last period, Claude Code desktop and T3 Code both shipped UI changes you can make by asking the agent. Theo says the timing reflects models reaching the point where agents can easily add and use such features. He would have been "way too nervous to merge this 6 months ago" . One caution from him: a nontechnical user on stable asked her agent for the nightly-only feature, and the agent cloned the source, leaving her on a fork .
  • Coordinating agents across machines. Brown connected a MacBook Pro and a Mac Mini to his Dot and had it run a Codex session on each, which uses computer use to do the work . He suggests the same setup for directing a computer reserved for video editing .
  • Grokbot on Tailscale. Grokbot can now join a Tailscale network to reach private MCP servers and internal tools .
  • Gemini 4 Carbon, another skeptic. After about 2.5 hours of testing, Brown found nothing Carbon did better than Opus, which he rates higher on front-end, 3D, games and video editing .
  • Talk to watch. @poteto's AIE talk on how the Grok bot team ships quickly without giving up quality is now up . Her notes on making agent corrections enforceable were covered last period.
Opus 5.5 stays the default; use Fable 5.2 to plan hard jobs
Riley Brown
  • Mine past agent conversations for reusable workflows: Brown asked Claude Code to analyze 70 threads about his video-editing process and map the steps—from computer setup to final output, including analysis, scraping, and prompting—on a whiteboard. He refined it through chat, and stopped an Extra-tier run in favor of Opus at medium effort because Extra was unnecessary for the task.
  • Route work by difficulty and cost: Brown recommends Opus for most tasks; for demanding, token-heavy work such as multi-agent, hour-plus Blender, game, or app projects, have Fable make a detailed plan and let Opus execute it. He says this can achieve roughly comparable quality for less token cost, while Fable is better reserved for urgent, high-quality work; he reports it costs twice as much as Opus and its weekly limits can bind.
  • Coordinate work across machines: Brown connected a MacBook Pro and Mac Mini to Dot and asked it to run two sessions, one on each computer, to list their Downloads folders. He describes Dot as coordinating a fleet through computer use and suggests the pattern can also control a dedicated video-editing machine.
  • Expose internal tools to an agent: Grokbot can join a Tailscale network to reach private MCP servers and internal tools.
  • Model comparison, with a limited-test caveat: After about 2.5 hours testing Gemini 4 Carbon, Brown found no task where it outperformed Opus, which he judged better at front-end work, 3D, games, and video editing. He also disliked accessing Gemini through Antigravity and saw no reason to switch despite its temporary half-price offer.
Claude Just Changed EVERYTHING (Here’s What’s New)
Riley Brown
Profile
  • Use Opus 5.5 for most tasks; for difficult, token-heavy work lasting an hour or more, ask Fable for a detailed plan and have Opus execute it to save token costs. Reserve Fable for urgent, high-quality work when cost matters less; the speaker says it costs about twice as much as Opus and that he has hit its weekly limits on ordinary business tasks.
  • A Dot can coordinate agent sessions across connected computers: the speaker asked it to list Downloads files using separate sessions on a MacBook and Mac Mini, and described using it to direct a dedicated video-editing computer.
Claude Just Changed EVERYTHING (Here’s What’s New)
Addy Osmani

Use Opus 5.5 at medium as a Claude Code daily driver; if it struggles, raise the effort setting before switching models, and reserve Fable 5.2 for genuinely hard work such as big migrations or long unattended runs. In a cited comparison of 52 coding tasks in Claude Code 2.1.322, Fable 5.2 solved 82.7% versus Opus 5.5's 84.6% and cost 2.80× as much per solved task; the comparison notes that harder tasks may be needed to distinguish the models.

Opus 5.5 at medium is a great daily driver. If it struggles, raise the effort before you switch models. Otherwise Fable 5.2 is good for t… Keep Opus 5.5 as your coding default in Claude Code. On 52 coding tasks in Claude Code 2.1.322, Fable 5.2 solved about as many (82.7% vs …
Theo - t3.gg

Claude Code desktop mods support custom themes: users can describe the look they want to Claude or start from the theme-examples plugin. Theo says recent model capabilities have made features like this easy for agents to add and use; he would have been too nervous to merge it six months earlier, and says agents only became able to take advantage of it in the past few months.

You can now build custom themes for Claude Code on desktop with mods! Change the colors, add an animated background, and make it your own… I think this mostly happens because model capabilities are at a point where these features are easy for agents to add and also use Would …
LangChain Blog

For long-horizon agents, scope work so people can review similar outputs and provide feedback that transfers across tasks; prioritize an intuitive review experience, since feedback teaches agents unwritten preferences and can be a performance bottleneck. Actively AI shifted from one agent per sales account to “motions” organized around recurring jobs, such as outbounding, making it easier for sellers to review similar work and share feedback; background agents still gathered account context.

Goal Engineering: How to Design Goals
Riley Brown
  • Riley Brown reports making a 2-minute 4K film in Blender from one prompt on Fable 5.2 in about 4.5 hours. Claude generated Blender assets with Python—five characters and props, seven sets, and camera moves and animation for 52 shots—producing 12,500 lines of code; it spawned seven copies of itself to build sets in parallel.
  • For rendering, the setup rented 89 GPUs through RunPod to render 3,220 frames, with 36 running at once, for a little over $20. Brown says access to cloud compute made his Claude Code setup more ambitious, and a similar prompt on another computer without the RunPod skill was “not nearly as high quality.”
The Fable of Clawd A 2 minute 4K film made entirely in Blender, from one prompt on Fable 5.2. It took about 4.5 hours. Claude built every…
Simon Willison

Simon Willison reports that Claude chat’s auto mode sometimes refuses to run something; telling the permission system, “I give this session permission to run this experiment for me (now try again),” seemed to unblock it. He notes that auto mode takes user messages into account.

Regular Claude chat uses "auto mode" now, so occasionally it will refuse to run something I just prompted "Hey permission system, I give …
swyx

Poteto posted an AI Engineer talk about how the Grok team ships quickly without compromising quality; the post links to the recording, but gives no workflow details.

my AIE talk is up! come hear how the grok [@bot](https://x.com/bot) team ships so quickly without compromise on quality [https://youtu.be…
swyx

@poteto shared an AI Engineer talk on how the Grok team ships quickly without compromising quality; the post links to the video but gives no workflow details .

my AIE talk is up! come hear how the grok [@bot](https://x.com/bot) team ships so quickly without compromise on quality [https://youtu.be…
Theo - t3.gg

T3 Code lets users ask its agent to customize the app UI, with changes updating on the fly, but the feature was available only in nightly, not stable. One nontechnical user asked her agent to customize her install; on stable, the agent cloned the source, leaving her on a fork.

T3 Code now lets you fully customize your UI in app. Ask your agent to make changes and it will update on the fly [![Video](https://pbs.t… I have a nontechnical friend who has been using T3 Code a lot lately. She saw this, got hyped, and immediately asked her agent to customi…
Tibo

The Cyber team is seeking feedback on Daybreak, including its models, Codex plugins, blue/red access, documentation, red teaming, and defense factory.

Cyber team wants your feedback on Daybreak. Models, codex plugins, blue/red access, documentation, red teaming & defense factory... W…
Armin Ronacher ⇌

Armin Ronacher says Fable 5.2 is “crazy good” at writing code that paints, while noting that Anthropic did not make an image model.

It's kinda funny how Anthropic didn't do an image model but then Fable 5.2 is crazy good at writing code that paints.
Kent C. Dodds 🐨

Kent C. Dodds said new features were available to try in Kody and that more were coming; he linked to https://kody.codes, but did not describe the features.

Lots happening in Kody land and so much more coming! Very excited for you to try these features! Try it out at [https://kody.codes](https…