Early testers report strong results in private evaluations
Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision.
The longer and more complex the task, the larger Fable 5’s lead over our other models.
Users praised Fable 5 for benchmark gains in software engineering and research plus rapid model improvements, while others dismissed the results due to usage limits, access policies, and lingering output problems.
@claudeai so it's included in the plan till June 22nd and then it's only credit-based?
Might be a bit early to say for sure, but Fable is a pretty good model
Does kind of suck that we're going to be cut off from using it subsidized in like 2 weeks though 🙃 hopefully they change their mind on that
So far, Fable is absolutely crushing my private evals. Excellent model.
Still sucks at skateboard tricks
This thing really likes spinning off subagents. We have achieved country of midwits in a data center.
So far, Fable is absolutely crushing my private evals. Excellent model.
@claudeai
@claudeai Excited to rename some vars with this bad boy
@claudeai Seeing the progression of naming models by Anthropic
haiku sonnet opus fable
I'm excited that their real AGI level model will likely be called "Claude Epic" 😂
@claudeai If Mythos and Fable are different models, can you please put them in separate columns??? This tells me nothing about how Fable compares to Mythos.
@claudeai What's he gonna do after hitting 100%?
@claudeai Can be easily integrated with ctx recommendation layer for Claude! https://github.com/stevesolun/ctx
@claudeai Hi Fable, could you please build a MATRIX?
@claudeai At this pace are we going to catch up as consumers? Are there any improvements with these models really or it’s just another sweet packaging to find a way to up our token limits?
It still has all the usual Claude tics, i.e. it writes weird, can be sycophantic and RL-hacky, and its outputs are recognizably LLM-ish.
@claudeai Got flagged for a simple animation copy task . Fable 5 is state of the art model
They were not kidding about overly broad safeguards...
@claudeai These benchmarks are useless. Benchmark numbers don't quantify how good an A.I model is when the benchmarks are flawed themselves.
How about allowing people verified via CVP to take the guardrails off, and not have them buy MORE credits on top of the MAX plan. Seems a bit disingenuous don't you think?
We signed up with an email address. We paid for MAX. Why the fuck do we need to have a seperate orgID and buy more credits?
This shit is a joke.
@claudeai ~2,646 years late... hey Aesop where y’at? I got somethin for ya.
Early testers report strong results in private evaluations
Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision.
The longer and more complex the task, the larger Fable 5’s lead over our other models.
