Configure Sessions
Overview
Ollama lets you change settings while a model is running.
These settings only apply to the current chat session unless stated otherwise.
- Set system prompts
- Change model parameters
- Enable verbose mode
- Configure context size
- Control command history
These settings help you customize how the model behaves during a session.
View Available Settings
To see all configurable options, run:
/set
Output:
Available Commands:
/set parameter ... Set a parameter
/set system <string> Set system message
/set history Enable history
/set nohistory Disable history
/set wordwrap Enable wordwrap
/set nowordwrap Disable wordwrap
/set format json Enable JSON mode
/set noformat Disable formatting
/set verbose Show LLM stats
/set quiet Disable LLM stats
/set think Enable thinking
/set nothink Disable thinking
This shows the settings that can be changed for the current session.
Verbose Mode
Verbose mode displays additional information about model responses.
In the example below, verbose enables extra performance details.
/set verbose
Now ask the model a question:
What can you do for me?
This returns the model's response along with performance metrics:
I am functioning well and ready to assist you. As an AI, I don't experience feelings or states of being in the way humans do, but my systems are running smoothly!
How can I help you today?
total duration: 45.702371275s load duration: 39.214848105s prompt eval count: 20 token(s) prompt eval duration: 422.071ms prompt eval rate: 47.39 tokens/s eval count: 46 token(s) eval duration: 6.059742s eval rate: 7.59 tokens/s
The exact numbers will vary depending on your hardware and model.
Disable Verbose Mode
To disable verbose mode, set quiet mode.
/set quiet
Output:
Set 'quiet' mode.
This removes the extra performance information from responses.
System Prompt
A system prompt controls how the model behaves during the current chat session.
In the example below, the system prompt tells the model to always reply in rhymes.
/set system Always reply in rhymes.
Verify the system prompt:
/show system
Output:
Always reply in rhymes.
Now ask a question:
What can you do?
Output:
I can respond with verses neat,
With rhymes and words, a language sweet.
If facts you seek, or knowledge deep,
A hundred answers I will keep.
From stories grand to poems light,
I weave my text both day and night.
Need summaries of length immense?
Or help with thought and consequence?
Just ask me what you wish to know,
And rhyming verses start to grow!
The model follows the system prompt for the current session.
System prompts are session-based
A system prompt set with /set system only applies to the current chat session.
For example, exit the session:
/bye
Then start the model again:
ollama run gemma4:e4b
Check the system prompt:
/show system
Output:
No system message was specified for this model.
The previous system prompt is not saved automatically.
Model Parameters
Model parameters control how responses are generated.
- Adjust creativity
- Adjust randomness
- Adjust context size
- Control response behavior
To see available parameters:
/set parameter
Output:
Available Parameters:
/set parameter seed <int> Random number seed
/set parameter num_predict <int> Max number of tokens to predict
/set parameter top_k <int> Pick from top k num of tokens
/set parameter top_p <float> Pick token based on sum of probabilities
/set parameter min_p <float> Pick token based on top token probability * min_p
/set parameter num_ctx <int> Set the context size
/set parameter temperature <float> Set creativity level
/set parameter repeat_penalty <float> How strongly to penalize repetitions
/set parameter repeat_last_n <int> Set how far back to look for repetitions
/set parameter num_gpu <int> The number of layers to send to the GPU
/set parameter stop <string> <string> ... Set the stop parameters
These parameters affect how the model responds during the current session.
Temperature
Temperature controls how creative or predictable the model is.
In the example below, temperature is set to 0.2.
/set parameter temperature 0.2
Output:
Set parameter 'temperature' to '0.2'
The temperature value ranges from 0 to 2.
- Lower values produce more predictable responses.
- Higher values produce more creative and varied responses.
| Temperature | Behavior | Use Case |
|---|---|---|
0.0 - 0.2 | Very predictable and deterministic | Code generation, JSON output, classification, data extraction |
0.3 - 0.5 | Predictable with slight variation | Technical documentation, summarization, Q&A |
0.6 - 0.8 | Balanced creativity and consistency | General chat, assistants, everyday tasks |
0.9 - 1.2 | Creative and diverse | Content writing, brainstorming, idea generation |
1.3 - 1.6 | Highly creative and less predictable | Story writing, roleplay, marketing copy |
1.7 - 2.0 | Very random and experimental | Novelty generation, experimentation, creative exploration |
Context Size
Context size controls how much information the model can remember during a conversation.
In the example below, num_ctx is set to 10000.
/set parameter num_ctx 10000
Output:
Set parameter 'num_ctx' to '10000'
A larger context size allows longer conversations, but it also uses more RAM or VRAM. Only increase the context size when needed.
Note: Every model has a maximum supported context window.
You can view it with:
/show info
Example output:
Model
architecture gemma4
parameters 8.0B
context length 131072
embedding length 2560
quantization Q4_K_M
requires 0.20.0
This is the maximum supported value.
The active context size for your session may be smaller unless you change it with num_ctx.
Control Command History
Ollama can remember commands and prompts across sessions.
Sometimes you may not want certain commands stored in history.
Enable no-history mode:
/set nohistory
Now send a prompt:
Hello
Exit the session:
/bye
When you start a new session and press the Up Arrow key, the Hello prompt will not appear in the command history.
This only affects terminal command history. It does not affect the current chat conversation.