I suppose a human could spend 10 seconds per Captcha, so they could do 360 per hour. Add some overhead for not being operating at peak performance every minute of every hour & call it 250. Let's say you can hire someone for $2, that works out to a bit over a penny per Captcha.
I don't think OpenAI has published pricing for GPT-4 Vision yet, but if we assume it's on par with GPT-4, and uses only 1000 of the 8000 possible tokens to process an image that's 3 cents per Captcha.
Doesn't seem completely unreasonable that at-scale humans may actually be cheaper than LLMs at this point. My mind is a little blown.
You'd be surprised, or perhaps horrified, by how cheap (self-proclaimed) human-based captcha solving services are.
If you just search for "captcha solving service" the first few results that come up offer 1000 solves of text-based captchas for <= $1 USD, (puzzle / JS browser challenge captchas are charged much higher).
Whether these are actually human based, or just impressive OCR services, it seems like they are still much more cost effective than GPT-4 is for now.
The way these work is usually presenting an existing captcha to another human who doesn’t even know they’re solving the captcha. For example, sites hosting pirated content serve fake captchas as a way to make money.
Given that it fails by hallucinating the structure of the challenge instead of refusing to solve a CAPTCHA, I doubt they've intentionally reduced the capability. Although the example in your sibling comment implies it should have enough information to do it.
This could be really interesting, we are currently using some custom bot detection techniques that also try to slow down / crash chrome headless and such, but it would be nice if we could just use this new Cloudflare service and stop worrying about maintaining it.
I have a few questions, though:
1. How does this work with mobile apps / single page applications APIs? I'm worried this would block legitimate users
2. Is it possible to enable/disable this feature on URLs patterns using page rules? I'm not seeing the option right now
3. Do blocked requests end up in the firewall logs? If so, how do we filter for it? Specific rule id?
4. Is there a way to enable it in log mode in order to evaluate the impact it would have?
Prima Assicurazioni (prima.it) | Milan, Italy | Full time | Onsite and italian language required | Backend/Frontend/Fullstack/DevOps | https://www.prima.it/carriera
We are a team of really smart people working in a very well funded startup trying to disrupt insurance in Italy by using technology as our main competitive advantage. We employ a micro service architecture (10s of them are in production atm), Docker and AWS.
Most of our micro services are written in Elixir, we also have some Ruby, Haskell, Python, Go and Rust in production besides a legacy Symfony 2 application, while on the front end side we are mainly using Elm.
We have a lot of automation in place, we run a pretty comprehensive test suite on Drone CI at every push and every developer can spin up QA environments that mirror our entire stack for a given feature branch (this is made possible by having all of our infrastructure managed through CloudFormation templates). All of this allows us to confidently deploy to production multiple times per day.
Our stack:
AWS (a lot of services), CloudFlare
OS: Amazon Linux
EC2 instance number: from ~30 to ~100: we scale automatically a lot of times during the day and treat our infrastructure as immutable
Infrastructure: CloudFormation + scripts
CD: Drone CI, Docker (dev -> qa -> staging -> production)
Container orchestration: AWS ECS
DB: Aurora MySQL and PostgreSQL, Redshift, ElastiCache Redis, DynamoDB
Monitoring: ELK, DataDog, New Relic, CloudWatch
Team organization: small cross functional agile teams (every team has at least one person for all of these roles: backend engineer, frontend engineer, web designer, qa engineer, devops engineer)
Backend languages: Elixir, PHP (we're phasing it out long term), Haskell, Python, Rust, Ruby, Go
Frontend languages: Elm, Javascript
Stuff that we like:
Micro service oriented architecture
Functional reactive programming
Event sourcing (CQRS)
Actor model
Agile (Scrum)
Domain-driven Design (DDD)
If you're interested or just want some more info feel free to drop me an email: andrea.usuelli@prima.it
Prima Assicurazioni (prima.it) | Milan, Italy | Full time | Onsite and italian language required | Backend/Frontend/Fullstack/DevOps | https://www.prima.it/carriera
We are a team of really smart people working in a very well funded startup trying to disrupt insurance in Italy by using technology as our main competitive advantage.
We employ a micro service architecture (10s of them are in production atm), Docker and AWS.
Most of our micro services are written in Elixir, we also have some Ruby, Haskell, Python and Rust in production besides a legacy Symfony 2 application, while on the front end side we are mainly using Elm.
We have a lot of automation in place, we run a pretty comprehensive test suite on Drone CI at every push and every developer can spin up QA environments that mirror our entire stack for a given feature branch (this is made possible by having all of our infrastructure managed through CloudFormation templates). All of this allows us to confidently deploy to production multiple times per day.
Our stack:
AWS (a lot of services), CloudFlare
OS: Amazon Linux
EC2 instance number: from ~30 to ~100: we scale automatically a lot of times during the day and treat our infrastructure as immutable
Infrastructure: CloudFormation + scripts
CD: Drone CI, Docker (dev -> qa -> staging -> production)
Container orchestration: AWS ECS
DB: Aurora MySQL and PostgreSQL, Redshift, ElastiCache Redis, DynamoDB
Monitoring: ELK, DataDog, New Relic, CloudWatch
Team organization: small cross functional agile teams (every team has at least one person for all of these roles: backend engineer, frontend engineer, web designer, qa engineer, devops engineer)
Backend languages: Elixir, PHP (we're phasing it out long term), Haskell, Python, Rust, Ruby
Frontend languages: Elm, Javascript
If you're interested or just want some more info feel free to email me directly at andrea.usuelli@prima.it
Prima Assicurazioni (prima.it) | Milan, Italy | Full time | Onsite and italian language required | Backend/Frontend/Fullstack/DevOps | https://www.prima.it/carriera
We're a team of really smart people working in a very well funded startup trying to disrupt insurance in Italy by using technology as our main competitive advantage.
We employ a micro service architecture (10s of them are in production atm), Docker and AWS.
Most of our micro services are written in Elixir (our language of choice for all the new stuff), we also have some Ruby, Haskell, Python and Rust in production besides a legacy Symfony 2 application.
On the front end side we're mainly using Elm.
We have a lot of automation in place, we run a pretty comprehensive test suite on Drone CI at every push and every developer can spin up QA environments that mirror our entire stack for a given feature branch (this is made possible by having all of our infrastructure managed through CloudFormation templates).
All of this allows us to confidently deploy to production multiple times per day.
Our stack:
AWS (a lot of services), CloudFlare
OS: Amazon Linux
EC2 instance number: from ~30 to ~100: we scale automatically a lot of times during the day and treat our infrastructure as immutable
Infrastructure: CloudFormation + scripts
CD: Drone CI, Docker (dev -> qa -> staging -> production)
Container orchestration: AWS ECS
DB: Aurora MySQL and PostgreSQL, Redshift, ElastiCache Redis, DynamoDB
Monitoring: ELK, DataDog, New Relic, CloudWatch
Team organization: small cross functional agile teams (every team has at least one person for all of these roles: backend engineer, frontend engineer, web designer, qa engineer, devops engineer)
Backend languages: Elixir, PHP (we're phasing it out long term), Haskell, Python, Rust, Ruby
Frontend languages: Elm, Javascript
If you're interested or just want some more info feel free to email me directly at andrea.usuelli@prima.it
I created it because I was annoyed with the lack of notifications provided by GitHub for some events like new people following you or starring/forking your projects.
A lot of people are now using it and that makes me happy even though I'm losing money by keeping it online.
I'm the author, feel free to ask me any question, I'm currently in the (albeit very slow) process of rewriting the whole project in Elixir. I'm also planning a few new features like organization notifications and better stats.
