I'm running through the same issue. It seems like the entire benchmarks package was removed from gym few days ago

openai/gym@4c460ba#diff-170d6bbf20fa02890840c39b9fbb66a0

As a temporary solution I can suggest a manual downgrade of gym to 0.9.5

pip uninstall gym
pip install gym==0.9.5

Dukeimg = I can confirm, this solved my problem. THANK YOU. I'll leave it as open because it's a major problem for the universe team.

Also - the pip install gym==0.9.5 didn't for for me (it installed the newest version anyway) so I grabbed the 0.9.5 version from here.

pip uninstall gym
pip install gym==0.9.5

It works.

Thanks so much for this Issue! I've been running around for the last two days trying to find a solution to all the errors this caused for me. I would suggest that you add this fix to the README

pip uninstall gym
pip install gym==0.9.5

Really It works.Thanks

when i use only gym then there is no error.
and when use gym and universe both then error occur.
I install universe and docker properly as per given instruction on github openai/universe page.
THE ERROR I GOT
shashank@shashank-HP-Notebook:~/Documents/Policy_Gradients_to_beat_Pong-master$ sudo python3 test.py
[2018-03-12 21:56:33,886] Making new env: flashgames.DuskDrive-v0
[2018-03-12 21:56:33,895] Writing logs to file: /tmp/universe-18074.log
Traceback (most recent call last):
File "/home/shashank/.local/lib/python3.6/site-packages/docker/client.py", line 87, in _raise_for_status
response.raise_for_status()
File "/usr/local/lib/python3.6/dist-packages/requests/models.py", line 935, in raise_for_status
raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 404 Client Error: Not Found for url: http+unix://var/run/docker.sock/vauto/containers/json?limit=-1&all=0&size=0&trunc_cmd=1

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "test.py", line 5, in
env.configure(remotes=1) # automatically creates a local docker container
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/wrappers/timer.py", line 14, in configure
self.env.configure(**kwargs)
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/wrappers/render.py", line 21, in configure
self.env.configure(**kwargs)
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/wrappers/throttle.py", line 32, in configure
self.env.configure(**kwargs)
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/envs/vnc_env.py", line 199, in configure
use_recorder_ports=record,
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/remotes/build.py", line 19, in build
n=n,
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/remotes/docker_remote.py", line 44, in init
self._assigner = PortAssigner(reuse=reuse)
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/remotes/docker_remote.py", line 165, in init
self._refresh_ports()
File "/home/shashank/.local/lib/python3.6/site-packages/universe/universe/universe/remotes/docker_remote.py", line 169, in _refresh_ports
for container in self.client.containers():
File "/home/shashank/.local/lib/python3.6/site-packages/docker/client.py", line 493, in containers
res = self._result(self._get(u, params=params), True)
File "/home/shashank/.local/lib/python3.6/site-packages/docker/client.py", line 93, in _result
self._raise_for_status(response)
File "/home/shashank/.local/lib/python3.6/site-packages/docker/client.py", line 89, in _raise_for_status
raise errors.APIError(e, response, explanation=explanation)
docker.errors.APIError: 404 Client Error: Not Found ("b'{"message":"page not found"}'")

PLEASE SOMEONE HELP.

This is a "the blind leading the blind" answer - to the extent I'm wrong PLEASE correct me.

First, I noticed that you used python 3.6 when universe only supports 3.5. Are you trying to learn to program AI? If so, I recommend...

1. A lot of OpenAI uses - if not Tensorflow - it uses the syntax of Tensorflow, Google's open source AI suite. The first two places to start is the python programming language if you need a crash course and then Tensorflow. Then install gym and baselines.

2. This guide helps set up gym and baselines. Be sure to use gym 9.5 if you're using universe. Honestly, I recommend not using universe for now. Universe seems to be a time-independent version of gym - if your program runs at 30 frames per second, gym expects to check in thirty times. Universe can roll without expecting that level of precision.

3. Install openai baselines. They're the AI suite. For example, copy baselines/deepq/experiments/train_carpole.py and enjoy_carpole.py to your coding directory. Use the first agent to train the AI quickly with the gui turned off, and the second agent to view what was trained. Congrats! You're using AI (with the cartpole environment).

@Accessant Thanks for help.