4 ms·

hey guys - Walden here, one of the founders. Excited to have you try out Devin. Reach out here if you have any questions!

adamgordonbell 2y ago
I'm excited to try it. I use aider quite a bit and tried opendevin at some point. What is the pricing story? Can I use it as side project dev or is the target enterprise customers only / mainly?
- waldenyan20 2y ago
  we have plenty of small, early stage teams that use Devin but it's optimized designed to fit into a team's workflow. you can of course give it a try and see if it's a good fit for your projects!
  - anticensor 2y ago
    Any estimates regarding when the personal tier ($50/month+credits) will resume accepting signups?
k2xl 2y ago
What model does it use under the hood? How much context window does it load when it is solving tasks? How does it determine which files to load into context?
- anticensor 2y ago
  It's a finetuned version of o1-preview sized distillation of o1-pro if I remember correctly, with an Azure Ubuntu VM with writable filesystem and internet access.
thekhatribharat 2y ago
How does one estimate the number of ACUs required to finish a task?
- waldenyan20 2y ago
  a helpful benchmark is that a typical frontend task is about 1-2 ACUs, but really depends on the complexity of the task
- anticensor 2y ago
  It spends about 2 to 10 ACU per hour in the small VM, and ten times as much on the large one. No credits spent during sleep and "waiting for response" time as far as I observed.
swyx 2y ago
as someone who has been trying you guys out for the past 8 months... you need a speed lever. default devin is way too slow for me :/ i asked scott for a "demo mode" first time we met
- waldenyan20 2y ago
  latest update is around 3-4x faster than it was back in Apr but we are working on making it much faster still!
  - kordlessagain 2y ago
    How is that done?
- neural_thing 2y ago
  Devin got a lot faster for me recently, made it a lot more enjoyable to use
JTyQZSnP3cQGa8B 2y ago
Why don't any LLM show examples of C++ applications? I have yet to see a tool like that which I would be happy to use at work.
- elashri 2y ago
  Or CUDA code, as this will be somehow ironic given that LLMs inference engines and training are CUDA code in some way.
  - anticensor 2y ago
    It can do that too, I tried that too.
- anticensor 2y ago
  I tried it with C and C++ code, it can do them but not very well.
  - menaerus 2y ago
    How large the repositories are that it can "reason" about?
    - anticensor 2y ago
      It's smart enough to figure out the relevant part to change once it scans the codebase. It could do Linux kernel if not for the Linus' policy.
      - menaerus 2y ago
        Thanks. For example, if I feed it with the 10 MLoC repository, how long does it take before it can start working through the problem? Would it really work well with the mixture of at least C and assembly which you implied it would with Linux kernel example?
papichulo2023 2y ago
Are you a human founder?
- waldenyan20 2y ago
  very much so!
  - xena 2y ago
    Can you upload a picture on the company domain with your face, and holding a piece of paper containing your name, the date, the time, and the current bitcoin block number? That would make us more likely to believe you are properly human.
mrieck 2y ago
I'd try it out if you allowed paying $50 for some credits instead of requiring subscription. Even if that version is limited to only editing public Github repos. $500 to see how well it works is too much.
- anticensor 2y ago
  $50/month in the subscription price in the personal tier (currently not accepting new users) includes 50 credits per month in it, and $500/month teams tier includes 250 credits per month in it. This is what I see with my current user and when I try to sign up as a new user respectively. I'd like to see that $50/month tier reopened to subscribers, and a $0/month+credits tier added (1 concurrent active session only, constrained to small VM spec with immutable rootfs (regular devin VMs have writable rootfs), no automatic knowledge generation, no snapshots, though playbooks allowed). > Even if that version is limited to only editing public Github repos Not possible to constrain like that with the current Devin architecture.
badFEengineer 2y ago
The price seems reasonable, but my main hesitation is on data storage + third party providers- there doesn't seem to be much available information on: * will you store my code + train on workflows that Devin does for me? * are you piping data to other third party providers (i.e. anthropic, openAI)?
cloudking 2y ago
When crafting projects from scratch, does your system actually fix it's own errors? That seems to be the challenge with Cursor Agent in it's current form, it generates a bunch of code that has bugs and requires a lot of iteration.
thekevan 2y ago
Can you only use it with a $500 / month subscription? The word "try" is VERY different than the actual case, which is "pay for use". If the answer to the first line is yes, how do I request my email be deleted? I started to sign up but I am not a use case for $500 a month at the moment.
- anticensor 2y ago
  It's monthly subscription plus prepaid compute credits (called ACU in the UI).
yuppiemephisto 2y ago
Does it work with more obscure languages like Lean 4?
- anticensor 2y ago
  It can work with any language, as it interacts with the VM and can read compiler messages.
anticensor 2y ago
You should really add an option to spawn a VM with immutable rootfs, current VMs all have writable rootfs which cost a lot to run, immutable VMs could be much much cheaper to operate (possibly enabling free tiers even). Also to mention, "suggest knowledge" modal is broken (it silently ignores changes made if you edit the suggested knowledge). Another issue, sleep&snapshot system is still prone to race conditions in certain cases.
tsak 2y ago
Is it just me that finds it ironic that you're looking for software developers?
throw83288 2y ago
Not really product related: The current trajectory of LLMs/Agents, what is your career advice to someone in school for Computer Science right now?
Buttons840 2y ago
Hi Walden, my name is Devin and I don't like sharing a name with a product. Will you please consider changing the name? There is always the chance that someone named Devin will do something that gives your product a bad name. Perhaps some new scandal will involve someone named Devin or something. I'd also like you to imagine that a hot new erotic AI was named "Walden", and people said things like "I was talking with Walden last night" as a euphemism. How would that make you feel?
anticensor 2y ago
Hey, can you fix the issue where the editor times out and Devin gets stuck?
marcusverus 2y ago
How should I go about arranging a demo?

adamgordonbell 2y ago

I'm excited to try it. I use aider quite a bit and tried opendevin at some point. What is the pricing story? Can I use it as side project dev or is the target enterprise customers only / mainly?

waldenyan20 2y ago
we have plenty of small, early stage teams that use Devin but it's optimized designed to fit into a team's workflow. you can of course give it a try and see if it's a good fit for your projects!
- anticensor 2y ago
  Any estimates regarding when the personal tier ($50/month+credits) will resume accepting signups?

k2xl 2y ago

What model does it use under the hood? How much context window does it load when it is solving tasks? How does it determine which files to load into context?

anticensor 2y ago
It's a finetuned version of o1-preview sized distillation of o1-pro if I remember correctly, with an Azure Ubuntu VM with writable filesystem and internet access.

thekhatribharat 2y ago

How does one estimate the number of ACUs required to finish a task?

waldenyan20 2y ago
a helpful benchmark is that a typical frontend task is about 1-2 ACUs, but really depends on the complexity of the task
anticensor 2y ago
It spends about 2 to 10 ACU per hour in the small VM, and ten times as much on the large one. No credits spent during sleep and "waiting for response" time as far as I observed.

swyx 2y ago

as someone who has been trying you guys out for the past 8 months... you need a speed lever. default devin is way too slow for me :/ i asked scott for a "demo mode" first time we met

waldenyan20 2y ago
latest update is around 3-4x faster than it was back in Apr but we are working on making it much faster still!
- kordlessagain 2y ago
  How is that done?
neural_thing 2y ago
Devin got a lot faster for me recently, made it a lot more enjoyable to use

JTyQZSnP3cQGa8B 2y ago

Why don't any LLM show examples of C++ applications? I have yet to see a tool like that which I would be happy to use at work.

elashri 2y ago
Or CUDA code, as this will be somehow ironic given that LLMs inference engines and training are CUDA code in some way.
- anticensor 2y ago
  It can do that too, I tried that too.
anticensor 2y ago
I tried it with C and C++ code, it can do them but not very well.
- menaerus 2y ago
  How large the repositories are that it can "reason" about?
  - anticensor 2y ago
    It's smart enough to figure out the relevant part to change once it scans the codebase. It could do Linux kernel if not for the Linus' policy.
    - menaerus 2y ago
      Thanks. For example, if I feed it with the 10 MLoC repository, how long does it take before it can start working through the problem? Would it really work well with the mixture of at least C and assembly which you implied it would with Linux kernel example?

papichulo2023 2y ago

Are you a human founder?

waldenyan20 2y ago
very much so!
- xena 2y ago
  Can you upload a picture on the company domain with your face, and holding a piece of paper containing your name, the date, the time, and the current bitcoin block number? That would make us more likely to believe you are properly human.

mrieck 2y ago

I'd try it out if you allowed paying $50 for some credits instead of requiring subscription. Even if that version is limited to only editing public Github repos. $500 to see how well it works is too much.

anticensor 2y ago
$50/month in the subscription price in the personal tier (currently not accepting new users) includes 50 credits per month in it, and $500/month teams tier includes 250 credits per month in it. This is what I see with my current user and when I try to sign up as a new user respectively. I'd like to see that $50/month tier reopened to subscribers, and a $0/month+credits tier added (1 concurrent active session only, constrained to small VM spec with immutable rootfs (regular devin VMs have writable rootfs), no automatic knowledge generation, no snapshots, though playbooks allowed). > Even if that version is limited to only editing public Github repos Not possible to constrain like that with the current Devin architecture.

badFEengineer 2y ago

The price seems reasonable, but my main hesitation is on data storage + third party providers- there doesn't seem to be much available information on: * will you store my code + train on workflows that Devin does for me? * are you piping data to other third party providers (i.e. anthropic, openAI)?

cloudking 2y ago

When crafting projects from scratch, does your system actually fix it's own errors? That seems to be the challenge with Cursor Agent in it's current form, it generates a bunch of code that has bugs and requires a lot of iteration.

thekevan 2y ago

Can you only use it with a $500 / month subscription? The word "try" is VERY different than the actual case, which is "pay for use". If the answer to the first line is yes, how do I request my email be deleted? I started to sign up but I am not a use case for $500 a month at the moment.

anticensor 2y ago
It's monthly subscription plus prepaid compute credits (called ACU in the UI).

yuppiemephisto 2y ago

Does it work with more obscure languages like Lean 4?

anticensor 2y ago
It can work with any language, as it interacts with the VM and can read compiler messages.

anticensor 2y ago

You should really add an option to spawn a VM with immutable rootfs, current VMs all have writable rootfs which cost a lot to run, immutable VMs could be much much cheaper to operate (possibly enabling free tiers even). Also to mention, "suggest knowledge" modal is broken (it silently ignores changes made if you edit the suggested knowledge). Another issue, sleep&snapshot system is still prone to race conditions in certain cases.

tsak 2y ago

Is it just me that finds it ironic that you're looking for software developers?

throw83288 2y ago

Not really product related: The current trajectory of LLMs/Agents, what is your career advice to someone in school for Computer Science right now?

Buttons840 2y ago

Hi Walden, my name is Devin and I don't like sharing a name with a product. Will you please consider changing the name? There is always the chance that someone named Devin will do something that gives your product a bad name. Perhaps some new scandal will involve someone named Devin or something. I'd also like you to imagine that a hot new erotic AI was named "Walden", and people said things like "I was talking with Walden last night" as a euphemism. How would that make you feel?

anticensor 2y ago

Hey, can you fix the issue where the editor times out and Devin gets stuck?

marcusverus 2y ago

How should I go about arranging a demo?