Categories AI

Fable 5.1: Ben’s Bites Guide

Greetings, everyone!

As I mentioned on Tuesday, I’ve been on quite a spree with building projects lately.

To kick things off, I scraped the OpenRouter website to create a tool that continuously updates to reflect the current stats.

Additionally, I received news that remove.bg, which specializes in background removal from images, will be shutting down in December following its acquisition by Canva. Consequently, I created my own version of the tool using Fable 5.1.

Many users pointed out that native functionality to achieve this exists on Macs (I was unaware of that), but the truth is that remove.bg was a business, and it was acquired for considerable value. Such tools are widely utilized.

I’m not creating these tools for anyone in particular or for profit. I enjoy building and exploring new possibilities. This endeavor brings me joy and valuable learning experiences about the process itself.

Every new project sparks additional ideas.

In tomorrow’s session recap, I’ll share insights from at least one of my projects, in which I crawled through 105 million rows of public government data to examine how our tax dollars are allocated… I ended up utilizing 2 billion tokens! 😳

All my projects will be made open-source, so you’ll have the chance to clone and remix them.

Here’s a snapshot of my usage metrics, by the way:

This tool will also be added to my site for you to clone. 😊 Stay tuned for tomorrow’s update!

p.s. I’ve already reached my Fable weekly usage limit. 🥹

Ben’s Bites is supported by Adobe

If you already use ChatGPT to brainstorm, research, and solve problems, you can now progress your conversations with the Adobe plugin in ChatGPT. Just describe the outcome you want, and the plugin utilizes over 70 Adobe tools to create and edit images, videos, designs, and documents. Get started today.

Claude Fable 5.1 has been released. This is currently the best model available. While it may not represent a clear leap in quality, I’ve found it considerably faster and easier to interact with over the last 24 hours. The model includes a new system prompt that prevents it from repeating song lyrics or copyrighted logos and alters its response style. Some exciting applications being developed with Fable include:

Anthropic has also cut the cost of caching inputs for Fable by 75%, making API usage about 25% less expensive than before.

OpenAI’s Astra is also expected soon (rumors suggest today).

Additionally, two new models are on the horizon: Gemini 3.8 Flash and Muse Spark 1.3. Both models are impressive and boast similar performance and base pricing; however, the pressing question remains: who can subsidize more tokens? Flash 3.8 currently offers a 50% discount for the remainder of the year, while Meta provides a 90% reduction in exchange for your data.

Another new model from Meta, Muse Voice Transcribe, looks incredibly promising for transcription applications. I hope it becomes open-source!

Gemini models now possess the capability to identify key sections of a video worth watching. Rather than sampling video at a fixed frame rate, it navigates through frames, audio, and transcripts to pinpoint essential moments. This method uses 88% fewer tokens, reduces costs by up to 66%, and offers a slight improvement in benchmark accuracy.

Inworld has developed Realtime TTS-2 for the upcoming surge of consumer applications. This technology requires three components: text-based voice design, latency under 100ms at P99, and a single voice identity across over 200 languages. Discover how to give your AI applications a voice.*

Share Ben’s Bites

This tool will also be part of my website, available for cloning. 😊 Stay tuned for updates tomorrow.

In closing, I’ve reached my usage limit for Fable this week. 🥹

Ben’s Bites is made possible thanks to Adobe

If you already use ChatGPT for brainstorming, research, and problem-solving, you can now elevate those conversations with the Adobe plugin in ChatGPT. Just describe your desired outcome, and the plugin leverages over 70 Adobe tools to create and edit images, videos, designs, and documents. Get started today.

Claude Fable 5.1 is here! This is the top model available right now. While there wasn’t a dramatic leap in quality, it’s notably faster and easier to engage with, based on my experience over the last 24 hours. The new system prompt prohibits repeating song lyrics or copyrighted logos and adjusts its response style. Here’s what folks are developing using Fable:

Anthropic has also reduced the cost of caching inputs for Fable by 75%, making API usage about 25% cheaper than previously.

OpenAI’s Astra is expected to launch soon (rumors suggest today).

There are also two exciting new models: Gemini 3.8 Flash and Muse Spark 1.3. Both deliver remarkable performance at comparable pricing; the key question is who can subsidize more tokens? Flash 3.8 is currently discounted by 50% until the end of the year, while Meta is offering a staggering 90% discount in exchange for data.

There’s also a fascinating new model from Meta: Muse Voice Transcribe, which looks incredibly promising for transcription applications. I hope it becomes open-source.

Gemini models can now determine which segments of video are worth watching. Instead of using a fixed sampling rate, it scans across frames, audio, and transcripts to identify the crucial moments. This approach can reduce token usage by up to 88%, lower costs by as much as 66%, and offer slight improvements in benchmark accuracy.

Inworld has introduced Realtime TTS-2 for the forthcoming wave of consumer applications. This solution integrates text-based voice design, achieves sub-100ms latency at P99, and maintains a single voice identity across over 200 languages. Discover how to empower your AI applications with a voice..*

Share Ben’s Bites



* sponsors who make this newsletter possible 🙂
Interested in partnering with us for the next quarter?

Leave a Reply

您的邮箱地址不会被公开。 必填项已用 * 标注

You May Also Like

Ben’s Bites: Transform Your Ideas into Reality

Building Ben’s Bites: My Journey

Who Allowed the Agents In?