Claude Fable 5.1

“A few notes on Anthropic's new Claude Fable 5.1 - with Max thinking level I got the best SVG pelican I've had from any Anthropic model (at a hefty cost of $3.30!), which I then had it animate simonwillison.net/2026/Sep/1/c...”

“Developer here, 15 years with no AI. Tested both. Im building server authoritative multiplayer game with persistance, prediction and long context system. Vps servers ecc... Astra is like 80% of Fable 5.1 But, the problem is, Fable one-shot your task 100% accurate and spend 15% of your weekly limit. While, Astra need 2 sessions and multiple iteration to finish the task, so at the end u also spend 15/20% of your weekly limit. So if u just play with AI, chose Astra, if you build something serious chose Fable”

“I just used it to do a review of a ~100k SLOC codebase that Fable 5 / Opus 5 largely built, cost like $2 and caught some good stuff, but more importantly, it communicated very directly and was pretty light on bizarre metaphors. No "let me read the source before opining" type verbiage launched at me. Honestly night and day for me vs before.”

“Almost exclusively Fable at xhigh. I sometimes run out of tokens, but most weeks it's not a problem. The limiting factor is by far how fast I can push understanding of what was made into my brain, and I find that I waste time using Opus. It makes annoying mistakes, and it makes assumptions that are often wrong instead of actually verifying. And it can't write so I can understand it.”

“Le code de Fable est devant en terme de production. J'arrive a des niveaux de confiance très elevé avec fable 5.1. Auj il m'est impossible d'avoir la même confiance avec Astra, trop variable avec ses sorties, beaucoup d'erreurs de suivie d'archi, beaucoup de slop, problème de cosntance & suivie des instructions. Astra est top sur des one shot demo, 3d, computer use. Mais du code déja en production très galère.”

“I've used Fable for so much stuff. My experience has been that it can pretty much one-shot most of my complex problems, if I describe them clearly and provide a solid way for it to verify its work. I get punted down to Opus 5 occasionally (for security-adjacent things) but that's pretty rare.”

“Been using Fable 5.1 since the minute it was release. It's actually a fix to all the issues I had with Fable 5 and Opus 5. The model isn't lecturing me either.”

“Had early access to Claude Fable 5.1. Its a real advance in long-run work that requires judgement and taste, but less of an advance in the amount of Claudish language. Here is a game from Fable 5.1 with retro graphics where you run an accurate space ship inspired by FTL cold-watch-game.netlify.app”

“What you completely miss is the context size -- by default you're capped at ~258k tokens context on Astra vs Fable's ~1K. And if you mod your config to go above that, you'll see your usage limit get consumed at an insane rate. Because of this, for large code bases, Astra simply can't cope. Fine for tiny toy games, but any serious work? No contest.”

“I literally built that same functionality with Opus 4.8. GPT 6 Astra is good, but I still don't think it's as good as Fable 5.1 for most tasks. Pretty much all the practical examples I've seen, plus my own testing, point to Fable 5.1 still being the better model overall. I'd say GPT 6 Astra is better at computer use and 3D stuff though. I think it also depends on the harness, the Codex harness is simply bad, Claude Code seems more mature and better for complex tasks.”

“Yeah, not even a contraversial take. You are straight wrong. Astra did a better job. The car on Fable barely even looked like a car, the people were worse than MC and the effects were lame. That camera movement is a bug not a feature btw. Astra did a better job and took less time and money”

“> In some cases, though, its prose is denser than Claude Fable 5's: sentences run longer and there are fewer paragraph breaks that feels like they just blocked words like load-bearing but can't actually fix the real problem. The insane word slop density and run on sentences was the real reason it became annoying to work with claude, colored with way too many analogies and pointless linguistic comparisons.”

“Real case: 2 days ago, I spent $100 on Fable 5.1 in about 20 minutes. $100 Pro, grind on it for 7 hours, still have lots of grinding avaiable. Astra, done, Fable 5.1 grind on 1 problem, never fixed, same coding problem with Astra, done in 5 minutes.”

“Nope. Always failed within 2-3 prompts. The most basic REST service you can imagine. Cookies are signed, that's crypto, banned. Completely useless model.”

“I've been using Astra and Fable with Unreal Engine MCP for the past few days. Fable was spinning it's wheels on a physics function. I made a handoff, and Astra fixed it in one turn. I asked for an explanation from Astra, presented it to Fable, and Fable basically said "I ignored your suggestions at every turn because I believed I knew better, but had I listened to you, it would have pointed me in the right direction, which is what the solution was". Anthropic frustrates the hell out of me.”

“2 problems: your new watermark thing just makes me reluctant to use your product if anyone can use an api key to know if i used ai to write something important. Another problem is that , unlike open ai who gives the best models to their plus users, fable is locked behind a pro subscription and then i still need to buy usage credits to use it. Making fable really just so expensive.”

“37:20 - this part is so true. The thing ChatGPT/Astra has over Fable is that you feel like you can use more of it. With Fable there's always this feeling of having to 2nd guess your every move so you don't blow your limit before the week's over. Astra has more wiggle room.”

“I'm thinking about cancelling my Claude Code subscription. This is my experience - Claude Code + Fable 5 = Hit 5-hour limit in a hour. - Claude Code + Opus 5 Add a feature Run /code-review Finds 7 issues Fix 7 issues Run /code-review Finds 9 issues, including regression errors”

“can't be that big of a deal, Every 9b finetune these days claims to be "Fable 5 level".... i was being sarcastic. But a while later i gotta say apart from the subway scene (at a heavy price), and the skate game after prompted fixes, this is not very impressive (for what you would expect). Or: Is this better than fable 5? The one impressive result that just worked, took hours? Have they nerfed this?”

“ngl all of these outputs look much worse than the previous claude models.. They expect us to pay API prices for this crap xD hell nah ill wait for opus 5.1”