Theo-brainUnofficial. Not affiliated with Theo or t3.gg.
Sign inYou
1:16:51
0:00

Fable Vs Astra Debate Is Over

Thursday, Sep 10 · 1 h 17 min · about 10 min to read

Theo-brain watched all 1 h 17 min

Theo compares Fable 5.1 and Astra on science, 3D, computer use, design, code quality, agents and cost. Fable is steady, with better design, feel and more mergeable code. Astra is spiky, far ahead at computer use and 3D looks, and much cheaper per task. He uses Fable to land code and Astra to operate his computer, and would pick Astra if forced to choose one.

Skim itThe notes carry the conclusions; watch these stretches to see the demos and failures that convinced him.

Parts

Setup: the two models that matter

Theo frames the comparison from heavy daily use of both models.

I'm doing nearly a billion tokens a day across both of these models consistently. And I've seen their strengths and weaknesses in all sorts of different places, many of which have

Science benchmarks

Astra on its lowest setting beats Fable 5.1's best score on Anthropic's own chosen benchmark, at under a third of the price.

until you went to the GBT6 Astra launch because Astra on low gets a 54.3 and only costs $11, putting low Astra higher than Max Fable 5.1 at under a third the price for the first benchmark that

3D rendering vs. how things feel

Astra's 3D looks generations better, but Fable handles interaction, animation and camera feel far better.

3D with Astra is comically better than 3D with Fable until you start interacting with it. And this is where Fable still just absolutely moss. Fable is so much better at interaction in

Computer use

Computer use is one of the two biggest gaps, alongside 3D: Astra in Codex flies through tasks on your machine.

with Astra. The biggest gaps by far are the 3D rendering where Astra clears and computer use where Astra also clears. It is so much faster than 5.6 was for computer use. It's crazy. It just flies through things on your machine. Computer

Copywriting and video work

Both models write more readable text now, but Astra stuffs pointless all-caps subtitles into every interface it builds.

really annoying edge, which is that it loves to stuff all cap subtitles into everything it builds in UI. It just throws these subtitles everywhere and it is the worst. Possibly the silliest

Front-end design

Fable clearly wins on design, though both still need heavy pushing to produce good work.

work. Fable takes the cake here easy, significantly better, but both still frustrate me constantly, and good designs still require you sit there with a hammer and beat the out of the model. So, while there is a gap

Part 7

22:35 to 33:23

brain. Astra is the smartest model that acts like it's dumb. Astra quadruple confirms everything it's trying to do, especially in X high mode. Which means if the bug is outside of what Fable can grasp just from reading code and it's something deeper or harder to find, Astra will find it. It'll just take four times longer and burn way more tokens.

Mergeability and scope creep

Fable writes more mergeable code; Astra balloons small fixes into huge pull requests.

with Astra to make the smallest possible changes, it will try to, but it will quickly have its context get bloated a bit by its reminders from all these review agents that will send it off course and then you end up with a thousand line of code PR that should have been 50. If you're in the loop

Part 9

42:17 to 50:20

I was able to merge like 150 PRs fully yolo merged with these models and had only two regressions. The first thing I feel obligated to say is that both of those regressions came from Astra. The

Orchestration, steering and skills

Astra wins orchestration with swarms: sub-agents that message each other and report back, unlike Fable's planned-up-front steps.

Astra, however, can just spin up a bunch of sub aents that are doing whatever and then pass messages to them, let them pass messages to each other, and most importantly, it can receive updates from those sub agents and use that to fan things out to keep the whole swarm

Cost and subscription limits

Astra uses far fewer tokens per task, so it ends up much cheaper despite the same list prices.

efficient by far. Meanwhile, Fable 5.1 is the least by far. Astra did in 27K tokens what Fable 5.1 did in almost 80K tokens. Take it as you will. It's also

Which one to pick

Fable is steady and Astra is spiky; if forced to choose one, he would take Astra.

quite a lot. If I had to pick one, I would pick Astra and have it write DMs to me to Julia so that he could send off the prompts using Fable for me. And I know it puts me in a weird place because I will always find some way to get good code out of the anthropic models even if I can't do it directly. Astra has more

Tap any time to jump the video there.

Key takeaways

His split: Fable 5.1 when he is trying to land code, Astra when he wants the AI to operate his computer. He says the code gap is not as big as it sounds.

That said, Fable is what I reach for when I'm trying to land code and Astra is what I reach for when I'm trying to use my computer. The gap for code is not as huge as it might sound. The gap is

Fable 5.1's pull requests average two follow-up fixes between being filed and being merged. Astra's average six. He measures this as how many issues his review bots catch before the code ships.

done the numbers here. Fable 5.1 from PR filed to merged averages two additional follow-ups from when the filed PR exists to when the merge happens. Meanwhile, Aster's been averaging six for me. What

Fable is consistent, while Astra swings between brilliant and baffling. If bad model moments throw off your work, pay extra for Fable. Astra is much cheaper and can do things no other model can.

off so much. If you are easily agitated to the point where it affects your work, you should pay the extra money for Fable. You just should. It's a way more consistent, reliable model. Astra, however, is way cheaper. It can do things no other model can, and it

Both models have the same per-token prices, yet on the Artificial Analysis index a task costs $3.26 with Astra versus $7.60 with Fable 5.1. The difference is that Astra uses far fewer tokens.

GB6 Astro was $3.26 per task. Opus 5 almost $6. Fable 5.1 $7.60. That is a comical gap even though they are priced the same. And theoretically speaking, the input token cost is four

On the same $200-a-month tier, he feels he gets about four times more done with Astra in Codex than with Fable in Claude Code. One reason: Claude only lets him spend half his usage on Fable.

It's just like I would say it's like a 4xish gap between the two is how it feels to me. feel like I can get four times more done with Codeex on Astra than I can with Fable and Claude Code on the same $200 a month tiers. But then

Four skills here are new to you.

Prototype with Astra, polish with Fable New

What it is
For 3D or interactive projects, Astra builds the first version to get the look roughly right. Then Fable handles the detail work.
Why Theo cares
Astra's 3D looks far better, but he calls it a sledgehammer. Fable handles animation curves, camera movement and overall feel much more gently.
How to try it
  1. Ask Astra for a first prototype of the interactive piece, focused on how it looks.
  2. Hand the result to Fable to clean up animations, camera and movement.
  3. Judge the result by using it, not by screenshots.
For my silly 3D demos that I've been working on, I find that the best way to build them is to make the first prototype with Astra to get everything roughly looking and feeling how you wantish and then have Fable come in and clean it up and do all of the like detailoriented work. Astra, in my

Rescue a stuck Astra pull request with Fable New

What it is
When Astra is looping on a pull request (a proposed code change) without finishing it, stop it and give the pull request to Fable 5.1 to finish.
Why Theo cares
He calls this his favorite use of Fable 5.1, because Fable is better at landing code cleanly.
How to try it
  1. Stop the Astra run once it is clearly stuck in a loop.
  2. Give Fable the pull request and say: make this actually land, clean it up, throw away whatever doesn't belong.
  3. Allow it to start a fresh branch (a separate copy of the code) if that is easier.
thing to use Fable 5.1 for to take something that Astra has in a death loop that it just cannot get through. Stop it. Switch over to Fable 5.1, hand it the PR and say, "Hey, make this actually land. Clean it up. Throw away whatever doesn't belong. Make a new branch if that's easier. Make the changes we care about here land." Bale 5.1 lands code

Astra sub-agent swarms New

What it is
Astra can launch many helper agents (sub-agents) that message each other and report back. It uses their updates to keep the whole group pointed in the right direction.
Why Theo cares
Fable plans its sub-agents in fixed steps up front. He has never seen anything like Astra's approach, and it tracks 40 parallel agents on his TypeScript-to-Rust port.
How to try it
  1. Give Astra one large, well-defined goal and let it spin up its own sub-agents.
  2. Check in on its progress rather than micromanaging each agent.
  3. Watch your usage closely, since swarms burn tokens fast.
Astra, however, can just spin up a bunch of sub aents that are doing whatever and then pass messages to them, let them pass messages to each other, and most importantly, it can receive updates from those sub agents and use that to fan things out to keep the whole swarm
with all of these agents. And I have 40 running here in parallel. And it can actually keep track of all of them and work with all of them. Unbelievable.

Astra computer use in Codex New

What it is
Astra in OpenAI's Codex app can operate your Mac directly: clicking, opening apps and carrying out tasks in the background.
Why Theo cares
It often finishes tasks faster than he would himself. He bought a second Mac Mini just to let it run around the clock. Ben Davis uses it to set up Final Cut, ready for video editing.
How to try it
  1. Try Astra computer use in the Codex app on a Mac.
  2. Start with prep work, like opening and arranging what you need before a task, rather than the creative work itself.
  3. If it proves useful, give it a spare machine so it doesn't disrupt yours.
With GBD6 Astra, I've been using it so much that I got another Mac Mini just to let it run with computer use 247 to do all sorts of different tasks. It's so good at realworld computer use. It can do things much faster. understands them
cleared with everything you need. Benis had a surprising amount of success getting Astra to do this type of work, to get it to actually set up his editor in Final Cut to be ready to go to edit

Advice

Do not trust Astra to keep your existing UI during a big rewrite. In his ping.gg rewrite, his prompt said to reuse as much UI code as possible, and Astra kept none of it.

this reason alone, I would make a bigger gap in the rewriting capabilities that we were just discussing. Because if the thing that you're doing this bulk rewrite for has UI, you cannot trust that UI is going to be carried over. Isentence in my prompt. Port features over one at a time. Reuse as much UI code as possible. Reuse as much UI code as possible. Do you see any UI code shared between these two versions of

Write your skill files (instruction files for agents) by hand, or at least audit them. Agent-written skills that then feed into writing more skills land you in a slop loop fast.

in slop hell very fast. This helps avoid that. Astra still falls into it. Fable falls into it even more. Please write your skills by hand or at least audit them and make some nice changes.

Don't reach for Astra's fast mode. He says using it regularly means you are burning tokens and not using Astra for its strengths.

very careful of the alter button and in particular the fast button. Straight up, fast is not the way you should use this model. If you're using fast with Astro, you're probably using Astro wrong or you're just burning tokens for fun, which I understand. But like if you're

If you use computers professionally and earn over $100,000 a year, you or your employer should pay for the $200-a-month Codex plan. It makes your computer use itself more effective.

almost feel like as a heavy user of computers, the $200 C codeex plan almost feels essential because it if you use computers professionally and you make over $100,000 a year, you or your boss should be paying for the $200 a month plan just because it makes your ability to use your computer more effective. If

Hot takes

Letting AI edit your videos or audio today is like shipping an app built with GPT-3.

His reasoning: He compares current AI video editing to GPT-3's toy React demo: impressive-looking, but nowhere near usable for real work.

with AI right now. If you let AI edit your content for you, if you let AI do your audio for you, if you let AI manage your AV pipeline for you in these ways, you are doing the equivalent of shipping an app using GPT3. We are that far

For writing copy, both new models still suck. The original Kimi K2 non-thinking model writes better than anything since, though it is dumb and rude.

His reasoning: He still rewrites most of the copy these models produce for his web pages.

hot take that they both still suck and nothing has topped Kimmy K2. Not even K3, not even K2.5. The original K2 non-thinking model still writes the best of any model I've used. But it's also stupid as hell and very quick to be incredibly rude. It's not a model you

Cache-write pricing is absurd. He pays Anthropic more to store a model's state for five minutes than to run the GPUs.

His reasoning: Cache writes are over 60% of his Fable spend. Cheaper cache reads barely matter because reads were only about 3% of his costs.

than that? Cash rights. Cash write costs are over 60% of my costs with Fable. So while it has made cash reads hilariously cheap, it has also made cash rights hilariously expensive. I am spending more money asking Anthropic to save the state of the model on their servers for

By the end of the year he expects a top-20 indie game on Steam built entirely with AI.

His reasoning: Neither model can make a good game alone, but using both carefully lets you iterate to a decent one.

and potentially make a decent game I do think we are now at the point where we're going to start seeing real games where the vast majority of everything was created via AI like I would guess by the end of the year we'll have our first top 20 indie game on Steam that was built entirely with AI so now we're out

How Theo thought about this

When Astra botched a simple revert, he ran a controlled test. He pasted the exact same prompt into Fable, adding only the dev-server request. Fable found the culprit and fixed it in about five minutes, while Astra had needed five corrections.

exact same prompt to Fable right after because I was so annoyed. This is an exact copy paste. The only difference being I added the third prompt in the spin up a tail scale dev server. I just copy pasted and added it to the prompt. Five minutes later, the auto scroll is

To judge mergeability, he placed Astra on a scale with older models, not just against Fable 5.1. He puts Astra roughly at or slightly ahead of Fable 5. It sits much closer to Fable 5.1 than GPT-5.6 Soul ever was to Fable. That old gap was about 3x, which made Soul hard to justify for code users would see.

reason I drew this diagram is that if you are thinking the gap between Astra and Fable is as big as between Soul and Fable, you're just wrong. This gap was massive. This was like a 3x difference that made it hard for me to justify using soul if the code would ever hit users. GB6 Astra is much much closer toThis is why I've been struggling. I think Astra is roughly at, if not slightly ahead of where Fable 5 was. So,

Sign in to mark what you've watched and keep notes.

Have a question about this one? Ask about this video →

Theo Daily is unofficial and not affiliated with Theo or t3.gg. Written by Claude Opus 5.5 from each video's transcript.