1[
2{ type: install
3  message: <<EOM
4You installed ollama: the AI model runner.
5
6To run ollama, plese open 2 terminals.
71. In the first terminal, please run:
8   $ ollama serve
92. In the second terminal, please run:
10   $ ollama run gemma3
11   or
12   $ ollama run mistral
13
14This will download and run the specified AI model.
15You will be able to interact with it in plain English.
16
17Please see https://ollama.com/library for the list
18of all supported models.
19
20The command "ollama list" lists all models downloaded
21into your system.
22
23When the model fails to load into your GPU, please use
24the provided ollama-limit-gpu-layers script to create
25model flavors with different num_gpu parameters.
26
27ollama uses many gigabytes of disk space in your home directory,
28because advanced AI models are often very large.
29Please symlink ~/.ollama to a large disk if needed.
30
31Working examples:
32(1) Coding with the model gpt-oss:20b:
33  1. run 'OLLAMA_CONTEXT_LENGTH=65536 ollama serve'
34     or
35     setup and start the service 'ollama'
36  2. install claude-code and run:
37     ANTHROPIC_BASE_URL=http://localhost:11434 \
38     ANTHROPIC_AUTH_TOKEN=ollama \
39     ANTHROPIC_MODEL=gpt-oss:20b \
40     ANTHROPIC_DEFAULT_SONNET_MODEL=gpt-oss:20b \
41     ANTHROPIC_DEFAULT_OPUS_MODEL=gpt-oss:20b \
42     ANTHROPIC_DEFAULT_HAIKU_MODEL=gpt-oss:20b \
43     claude
44  3. Ask it to write some program.
45
46(2) Image generation with the model x/z-image-turbo:
47  1. run 'OLLAMA_CONTEXT_LENGTH=65536 ollama serve'
48     or
49     setup and start the service 'ollama'
50  2. install claude-code and run:
51     ollama run x/z-image-turbo {textual description of the desired image}
52
53There are also a lot of text-to-text models that you can chat with.
54
55EOM
56}
57]
58