feat: Add reasoning_effort validation for the main configuration file

# Conflicts:
#	src/repl/mod.rs

assets/agents/code-reviewer/README.md

@@ -16,7 +16,7 @@ agents while handling coordination and final reporting.
 ## Pro-Tip: Use an IDE MCP Server for Improved Performance
 Many modern IDEs now include MCP servers that let LLMs perform operations within the IDE itself and use IDE tools. Using
 an IDE's MCP server dramatically improves the performance of coding agents. So if you have an IDE, try adding that MCP
-server to your config (see the [MCP Server docs](../../../docs/function-calling/MCP-SERVERS.md) to see how to configure
+server to your config (see the [MCP Server docs](https://github.com/Dark-Alex-17/coyote/wiki/MCP-Servers) to see how to configure
 them), and modify the agent definition to look like this:
 
 ```yaml

assets/agents/file-reviewer/README.md

@@ -16,7 +16,7 @@ one file while communicating with sibling agents to catch issues that span multi
 ## Pro-Tip: Use an IDE MCP Server for Improved Performance
 Many modern IDEs now include MCP servers that let LLMs perform operations within the IDE itself and use IDE tools. Using
 an IDE's MCP server dramatically improves the performance of coding agents. So if you have an IDE, try adding that MCP
-server to your config (see the [MCP Server docs](../../../docs/function-calling/MCP-SERVERS.md) to see how to configure
+server to your config (see the [MCP Server docs](https://github.com/Dark-Alex-17/coyote/wiki/MCP-Servers) to see how to configure
 them), and modify the agent definition to look like this:
 
 ```yaml

config.agent.example.yaml

@@ -13,6 +13,8 @@
 model: openai:gpt-4o             # Specify the LLM to use
 temperature: null                # Set default temperature parameter, range (0, 1)
 top_p: null                      # Set default top-p parameter, with a range of (0, 1) or (0, 2) depending on the model
+reasoning_effort: null           # Reasoning effort level for models that support it (e.g. low, medium, high).
+                                 # Only valid when the agent's model declares reasoning_levels.
 agent_session: null              # Set a session to use when starting the agent. (e.g. temp, default); defaults to globally set agent_session
 name: <agent-name>               # Name of the agent, used in the UI and logs
 description: <description>       # Description of the agent, used in the UI

config.example.yaml

@@ -2,6 +2,8 @@
 model: openai:gpt-4o             # Specify the LLM to use
 temperature: null                # Set default temperature parameter (0, 1)
 top_p: null                      # Set default top-p parameter, with a range of (0, 1) or (0, 2) depending on the model
+reasoning_effort: null           # Reasoning effort level for models that support it (e.g. low, medium, high).
+                                 # Only valid when the active model declares reasoning_levels. See the Clients docs.
 
 # ---- Behavior ----
 stream: true                     # Controls whether to use the stream-style APIs when querying for completions from LLM clients.

config.role.example.md

@@ -8,6 +8,8 @@ name: <role-name>                     # The name of the role
 model: openai:gpt-4o                  # The model to use for this role
 temperature: 0.2                      # The temperature to use for this role when querying the model
 top_p: 0                              # The top_p to use for this role when querying the model
+reasoning_effort: null                # Reasoning effort level for models that support it (e.g. low, medium, high).
+                                      # Only valid when the role's model declares reasoning_levels.
 enabled_tools:                        # Tools to enable for this role. Accepts a YAML list (preferred)
   - fs_ls                             # or a comma-separated string (e.g. `enabled_tools: fs_ls,fs_cat`).
   - fs_cat                            # Use `all` to enable every visible tool.

graph.example.yaml

@@ -33,6 +33,8 @@ version: "1.0"                     # Graph schema version. Only "1.0" is accepte
 model: claude:claude-sonnet-4-6    # Default model for `llm` nodes that don't override it
 temperature: 0.0                   # Default sampling temperature for `llm` nodes
 top_p: null                        # Default sampling top-p for `llm` nodes
+reasoning_effort: null             # Default reasoning effort for `llm` nodes that don't override it.
+                                   # Only valid when the model declares reasoning_levels.
 
 global_tools:                      # Tool universe an `llm` node's `tools:` whitelist draws from
   - web_search_coyote.sh
@@ -394,6 +396,7 @@ nodes:
       - mcp:ddg-search         #   `mcp:<server>` includes that server's functions
     model: claude:claude-haiku-4-5  # Optional per-node model override
     temperature: 0.3           # Optional per-node sampling override
+    reasoning_effort: null     # Optional per-node reasoning effort override (e.g. low, medium, high)
     max_attempts: 2            # Retry count on transient errors only. Default 1.
     max_iterations: 10         # Tool-call-loop turn cap. Default 10.
     fallback: review           # Route here if all attempts fail

models.yaml

@@ -10,6 +10,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: gpt-5.6-terra
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -17,6 +19,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: gpt-5.6-luna
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -24,6 +28,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: gpt-5.5
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -31,6 +37,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: gpt-5.5-pro
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -38,6 +46,8 @@
       output_price: 180
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: high
     - name: gpt-5.4
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -45,6 +55,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: gpt-5.4-pro
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -52,6 +64,8 @@
       output_price: 180
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: gpt-5.4-mini
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -59,6 +73,8 @@
       output_price: 4.5
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: gpt-5.4-nano
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -66,6 +82,8 @@
       output_price: 1.25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: gpt-5.3-codex
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -73,6 +91,8 @@
       output_price: 14
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: chat-latest
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -87,6 +107,17 @@
       output_price: 14
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
+    - name: gpt-5.2-pro
+      max_input_tokens: 400000
+      max_output_tokens: 128000
+      input_price: 21
+      output_price: 168
+      supports_vision: true
+      supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: gpt-5.1
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -94,6 +125,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high]
+      default_reasoning_effort: none
     - name: gpt-5.1-chat-latest
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -101,6 +134,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high]
+      default_reasoning_effort: none
     - name: gpt-5
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -108,6 +143,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
     - name: gpt-5-chat-latest
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -115,6 +152,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
     - name: gpt-5-mini
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -172,6 +211,8 @@
       supports_vision: true
       supports_function_calling: true
       system_prompt_prefix: Formatting re-enabled
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: medium
       patch:
         body:
           max_tokens: null
@@ -285,18 +326,24 @@
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
     - name: gemini-3-flash-preview
       max_input_tokens: 1048576
       max_output_tokens: 65536
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-3.1-flash-lite
       max_input_tokens: 1048576
       max_output_tokens: 65536
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: minimal
     - name: gemini-3.1-pro-preview
       max_input_tokens: 1048576
       max_output_tokens: 65535
@@ -304,6 +351,8 @@
       output_price: 2.5
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-2.5-flash
       max_input_tokens: 1048576
       max_output_tokens: 65536
@@ -311,6 +360,8 @@
       output_price: 0
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: low
     - name: gemini-2.5-pro
       max_input_tokens: 1048576
       max_output_tokens: 65536
@@ -318,6 +369,8 @@
       output_price: 0
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-2.5-flash-lite
       max_input_tokens: 1000000
       max_output_tokens: 64000
@@ -329,10 +382,14 @@
       max_input_tokens: 1048576
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, high]
+      default_reasoning_level: high
     - name: gemini-3-flash-preview
       max_input_tokens: 1048576
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_level: high
     - name: gemma-3-27b-it
       max_input_tokens: 131072
       max_output_tokens: 8192
@@ -358,6 +415,8 @@
       output_price: 50
       supports_function_calling: true
       supports_vision: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-8
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -366,6 +425,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-7
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -374,6 +435,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -382,6 +445,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-6:thinking
       real_name: claude-opus-4-6
       max_input_tokens: 200000
@@ -406,6 +471,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-sonnet-4-6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -414,6 +481,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: claude-sonnet-4-6:thinking
       real_name: claude-sonnet-4-6
       max_input_tokens: 200000
@@ -856,18 +925,24 @@
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
     - name: gemini-3-flash-preview
       max_input_tokens: 1048576
       max_output_tokens: 65536
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-3.1-flash-lite
       max_input_tokens: 1048576
       max_output_tokens: 65536
       input_price: 0.2
       output_price: 1.5
       supports_function_calling: true
+      reasoning_levels: [minimal, high]
+      default_reasoning_effort: minimal
     - name: gemini-3.1-pro-preview
       max_input_tokens: 1048576
       max_output_tokens: 65536
@@ -875,6 +950,8 @@
       output_price: 12
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-2.5-flash
       max_input_tokens: 1048576
       max_output_tokens: 65535
@@ -882,6 +959,8 @@
       output_price: 2.5
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: medium
     - name: gemini-2.5-pro
       max_input_tokens: 1048576
       max_output_tokens: 65536
@@ -889,6 +968,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
     - name: gemini-2.5-flash-lite
       max_input_tokens: 1048576
       max_output_tokens: 65536
@@ -900,10 +981,14 @@
       max_input_tokens: 1048576
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, high]
+      default_reasoning_effort: high
     - name: gemini-3-flash-preview
       max_input_tokens: 1048576
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: high
     - name: claude-fable-5
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -912,6 +997,8 @@
       output_price: 50
       supports_function_calling: true
       supports_vision: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-8
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -920,6 +1007,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-7
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -928,6 +1017,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-opus-4-6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -959,6 +1050,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: claude-sonnet-4-6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -967,6 +1060,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: claude-sonnet-4-6:thinking
       real_name: claude-sonnet-4-6
       max_input_tokens: 200000
@@ -1099,6 +1194,8 @@
       output_price: 50
       supports_function_calling: true
       supports_vision: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-opus-4-8
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -1107,6 +1204,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-opus-4-7
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -1115,6 +1214,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-opus-4-6-v1
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -1123,6 +1224,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-opus-4-6-v1:thinking
       real_name: us.anthropic.claude-opus-4-6-v1
       max_input_tokens: 200000
@@ -1148,6 +1251,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-sonnet-4-6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -1156,6 +1261,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: us.anthropic.claude-sonnet-4-6:thinking
       real_name: us.anthropic.claude-sonnet-4-6
       max_input_tokens: 200000
@@ -1672,6 +1779,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.6-terra
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1679,6 +1788,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.6-luna
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1686,6 +1797,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh, max]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.5
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1693,6 +1806,8 @@
       output_price: 30
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.5-pro
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1700,6 +1815,8 @@
       output_price: 180
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: high
     - name: openai/gpt-5.4
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1707,6 +1824,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: openai/gpt-5.4-pro
       max_input_tokens: 1050000
       max_output_tokens: 128000
@@ -1714,6 +1833,8 @@
       output_price: 180
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.4-mini
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1721,6 +1842,8 @@
       output_price: 4.5
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: openai/gpt-5.4-nano
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1728,6 +1851,8 @@
       output_price: 1.25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
     - name: openai/gpt-5.3-codex
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1735,6 +1860,8 @@
       output_price: 14
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: openai/gpt-5.2
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1742,6 +1869,17 @@
       output_price: 14
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [none, low, medium, high, xhigh]
+      default_reasoning_effort: none
+    - name: openai/gpt-5.2-pro
+      max_input_tokens: 400000
+      max_output_tokens: 128000
+      input_price: 21
+      output_price: 168
+      supports_vision: true
+      supports_function_calling: true
+      reasoning_levels: [medium, high, xhigh]
+      default_reasoning_effort: medium
     - name: openai/gpt-5
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1749,6 +1887,8 @@
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
     - name: openai/gpt-5-mini
       max_input_tokens: 400000
       max_output_tokens: 128000
@@ -1786,18 +1926,67 @@
       input_price: 0.04
       output_price: 0.16
       supports_function_calling: true
+    - name: google/gemini-3.5-flash
+      max_input_tokens: 1048576
+      max_output_tokens: 65536
+      input_price: 0.2
+      output_price: 1.5
+      supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: medium
+    - name: google/gemini-3-flash-preview
+      max_input_tokens: 1048576
+      max_output_tokens: 65536
+      input_price: 0.2
+      output_price: 1.5
+      supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: high
+    - name: google/gemini-3.1-flash-lite
+      max_input_tokens: 1048576
+      max_output_tokens: 65536
+      input_price: 0.2
+      output_price: 1.5
+      supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_effort: minimal
+    - name: google/gemini-3.1-pro-preview
+      max_input_tokens: 1048576
+      max_output_tokens: 65535
+      input_price: 0.3
+      output_price: 2.5
+      supports_vision: true
+      supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
+    - name: google/gemini-3-pro-preview
+      max_input_tokens: 1048576
+      supports_vision: true
+      supports_function_calling: true
+      reasoning_levels: [low, high]
+      default_reasoning_level: high
+    - name: google/gemini-3-flash-preview
+      max_input_tokens: 1048576
+      supports_vision: true
+      supports_function_calling: true
+      reasoning_levels: [minimal, low, medium, high]
+      default_reasoning_level: high
     - name: google/gemini-2.5-flash
       max_input_tokens: 1048576
       input_price: 0.3
       output_price: 2.5
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: low
     - name: google/gemini-2.5-pro
       max_input_tokens: 1048576
       input_price: 1.25
       output_price: 10
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high]
+      default_reasoning_effort: high
     - name: google/gemini-2.5-flash-lite
       max_input_tokens: 1048576
       input_price: 0.3
@@ -1827,6 +2016,8 @@
       output_price: 50
       supports_function_calling: true
       supports_vision: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-opus-4-8
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -1835,6 +2026,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-opus-4-7
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -1843,6 +2036,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-opus-4.6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -1851,6 +2046,8 @@
       output_price: 25
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-sonnet-5
       max_input_tokens: 1000000
       max_output_tokens: 128000
@@ -1859,6 +2056,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, xhigh, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-sonnet-4.6
       max_input_tokens: 200000
       max_output_tokens: 8192
@@ -1867,6 +2066,8 @@
       output_price: 15
       supports_vision: true
       supports_function_calling: true
+      reasoning_levels: [low, medium, high, max]
+      default_reasoning_effort: high
     - name: anthropic/claude-opus-4.5
       max_input_tokens: 200000
       max_output_tokens: 8192

src/client/bedrock.rs

@@ -325,6 +325,7 @@ fn build_chat_completions_body(data: ChatCompletionsData, model: &Model) -> Resu
         mut messages,
         temperature,
         top_p,
+        reasoning_effort,
         functions,
         stream: _,
     } = data;
@@ -457,6 +458,9 @@ fn build_chat_completions_body(data: ChatCompletionsData, model: &Model) -> Resu
     if let Some(v) = top_p {
         body["inferenceConfig"]["topP"] = v.into();
     }
+    if let Some(v) = reasoning_effort {
+        body["additionalModelRequestFields"] = json!({ "output_config": { "effort": v } });
+    }
     if let Some(functions) = functions {
         let tools: Vec<_> = functions
             .iter()

src/client/claude.rs

@@ -251,6 +251,7 @@ pub fn claude_build_chat_completions_body(
         mut messages,
         temperature,
         top_p,
+        reasoning_effort,
         functions,
         stream,
     } = data;
@@ -369,6 +370,9 @@ pub fn claude_build_chat_completions_body(
     if let Some(v) = top_p {
         body["top_p"] = v.into();
     }
+    if let Some(v) = reasoning_effort {
+        body["output_config"] = json!({ "effort": v });
+    }
     if stream {
         body["stream"] = true.into();
     }

src/client/common.rs

@@ -286,6 +286,7 @@ pub struct ChatCompletionsData {
     pub messages: Vec<Message>,
     pub temperature: Option<f64>,
     pub top_p: Option<f64>,
+    pub reasoning_effort: Option<String>,
     pub functions: Option<Vec<FunctionDeclaration>>,
     pub stream: bool,
 }

src/client/model.rs

@@ -289,6 +289,14 @@ impl Model {
         }
         Ok(())
     }
+
+    pub fn reasoning_levels(&self) -> &[String] {
+        &self.data.reasoning_levels
+    }
+
+    pub fn default_reasoning_effort(&self) -> Option<&str> {
+        self.data.default_reasoning_effort.as_deref()
+    }
 }
 
 #[derive(Debug, Clone, Default, Serialize, Deserialize)]
@@ -316,6 +324,10 @@ pub struct ModelData {
     pub supports_vision: bool,
     #[serde(default, skip_serializing_if = "std::ops::Not::not")]
     pub supports_function_calling: bool,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub reasoning_levels: Vec<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub default_reasoning_effort: Option<String>,
     #[serde(default, skip_serializing_if = "std::ops::Not::not")]
     no_stream: bool,
     #[serde(default, skip_serializing_if = "std::ops::Not::not")]

src/client/openai.rs

@@ -356,6 +356,7 @@ pub fn openai_build_chat_completions_body(data: ChatCompletionsData, model: &Mod
         messages,
         temperature,
         top_p,
+        reasoning_effort,
         functions,
         stream,
     } = data;
@@ -454,6 +455,9 @@ pub fn openai_build_chat_completions_body(data: ChatCompletionsData, model: &Mod
     if let Some(v) = top_p {
         body["top_p"] = v.into();
     }
+    if let Some(v) = reasoning_effort {
+        body["reasoning_effort"] = v.into();
+    }
     if stream {
         body["stream"] = true.into();
     }
@@ -534,6 +538,7 @@ pub fn openai_build_responses_body(data: ChatCompletionsData, model: &Model) ->
         messages,
         temperature,
         top_p,
+        reasoning_effort,
         functions,
         stream,
     } = data;
@@ -590,6 +595,9 @@ pub fn openai_build_responses_body(data: ChatCompletionsData, model: &Model) ->
     if let Some(v) = top_p {
         body["top_p"] = v.into();
     }
+    if let Some(v) = reasoning_effort {
+        body["reasoning"] = json!({ "effort": v });
+    }
     if stream {
         body["stream"] = true.into();
     }

src/client/vertexai.rs

@@ -334,6 +334,7 @@ pub fn gemini_build_chat_completions_body(
         mut messages,
         temperature,
         top_p,
+        reasoning_effort,
         functions,
         stream: _,
     } = data;
@@ -426,6 +427,9 @@ pub fn gemini_build_chat_completions_body(
     if let Some(v) = top_p {
         body["generationConfig"]["topP"] = v.into();
     }
+    if let Some(v) = reasoning_effort {
+        body["generation_config"]["thinking_level"] = v.into();
+    }
 
     if let Some(functions) = functions {
         // Gemini doesn't support functions with parameters that have empty properties, so we need to patch it.

src/config/agent.rs

@@ -575,6 +575,10 @@ impl RoleLike for Agent {
         self.config.top_p
     }
 
+    fn reasoning_effort(&self) -> Option<String> {
+        self.config.reasoning_effort.clone()
+    }
+
     fn enabled_tools(&self) -> Option<Vec<String>> {
         None
     }
@@ -596,6 +600,10 @@ impl RoleLike for Agent {
         self.config.top_p = value;
     }
 
+    fn set_reasoning_effort(&mut self, value: Option<String>) {
+        self.config.reasoning_effort = value;
+    }
+
     fn set_enabled_tools(&mut self, value: Option<Vec<String>>) {
         match value {
             Some(tools) => {
@@ -637,6 +645,8 @@ pub struct AgentConfig {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub top_p: Option<f64>,
     #[serde(skip_serializing_if = "Option::is_none")]
+    pub reasoning_effort: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
     pub agent_session: Option<String>,
     #[serde(default)]
     pub auto_continue: bool,
@@ -732,6 +742,7 @@ impl AgentConfig {
             model_id: graph.model.clone(),
             temperature: graph.temperature,
             top_p: graph.top_p,
+            reasoning_effort: graph.reasoning_effort.clone(),
             description: graph.description.clone(),
             global_tools: graph.global_tools.clone(),
             mcp_servers: graph.mcp_servers.clone(),
@@ -766,6 +777,9 @@ impl AgentConfig {
         if let Some(v) = read_env_value::<f64>(&with_prefix("top_p")) {
             self.top_p = v;
         }
+        if let Some(v) = read_env_value::<String>(&with_prefix("reasoning_effort")) {
+            self.reasoning_effort = v;
+        }
         if let Ok(v) = env::var(with_prefix("global_tools"))
             && let Ok(v) = serde_json::from_str(&v)
         {

src/config/app_config.rs

@@ -1,4 +1,4 @@
-use crate::client::{ClientConfig, list_models};
+use crate::client::{ClientConfig, Model, ModelType, list_models};
 use crate::render::{MarkdownRender, RenderOptions};
 use crate::utils::{IS_STDOUT_TERMINAL, NO_COLOR, decode_bin, get_env_name};
 
@@ -21,6 +21,7 @@ pub struct AppConfig {
     pub model_id: String,
     pub temperature: Option<f64>,
     pub top_p: Option<f64>,
+    pub reasoning_effort: Option<String>,
 
     pub dry_run: bool,
     pub stream: bool,
@@ -100,6 +101,7 @@ impl Default for AppConfig {
             model_id: Default::default(),
             temperature: None,
             top_p: None,
+            reasoning_effort: None,
 
             dry_run: false,
             stream: true,
@@ -177,6 +179,7 @@ impl AppConfig {
             model_id: config.model_id,
             temperature: config.temperature,
             top_p: config.top_p,
+            reasoning_effort: None,
 
             dry_run: config.dry_run,
             stream: config.stream,
@@ -253,6 +256,7 @@ impl AppConfig {
         app_config.setup_document_loaders();
         app_config.setup_user_agent();
         app_config.resolve_model()?;
+        app_config.validate_reasoning_effort()?;
         Ok(app_config)
     }
 
@@ -273,6 +277,31 @@ impl AppConfig {
         Ok(())
     }
 
+    fn validate_reasoning_effort(&self) -> Result<()> {
+        let Some(ref effort) = self.reasoning_effort else {
+            return Ok(());
+        };
+        let model = Model::retrieve_model(self, &self.model_id, ModelType::Chat)?;
+        let levels = model.reasoning_levels();
+
+        if levels.is_empty() {
+            bail!(
+                "reasoning_effort '{}' is configured but the model does not support reasoning effort",
+                effort
+            );
+        }
+
+        if !levels.iter().any(|l| l == effort) {
+            bail!(
+                "reasoning_effort '{}' is not valid for the model. Supported levels: {}",
+                effort,
+                levels.join(", ")
+            );
+        }
+
+        Ok(())
+    }
+
     pub fn resolve_model(&mut self) -> Result<()> {
         if self.model_id.is_empty() {
             let models = list_models(self, crate::client::ModelType::Chat);
@@ -426,6 +455,9 @@ impl AppConfig {
         if let Some(v) = super::read_env_value::<f64>(&get_env_name("top_p")) {
             self.top_p = v;
         }
+        if let Some(v) = super::read_env_value::<String>(&get_env_name("reasoning_effort")) {
+            self.reasoning_effort = v;
+        }
 
         if let Some(Some(v)) = super::read_env_bool(&get_env_name("dry_run")) {
             self.dry_run = v;

src/config/input.rs

@@ -253,6 +253,10 @@ impl Input {
         patch_messages(&mut messages, model);
         model.guard_max_input_tokens(&messages)?;
         let (temperature, top_p) = (self.role().temperature(), self.role().top_p());
+        let reasoning_effort = self
+            .role()
+            .reasoning_effort()
+            .or_else(|| model.default_reasoning_effort().map(|s| s.to_string()));
         let functions = if model.supports_function_calling() {
             let fns = self.functions.clone();
             if let Some(vec) = &fns {
@@ -268,6 +272,7 @@ impl Input {
             messages,
             temperature,
             top_p,
+            reasoning_effort,
             functions,
             stream,
         })

src/config/request_context.rs

@@ -969,6 +969,16 @@ impl RequestContext {
         }
     }
 
+    pub fn set_reasoning_effort_on_role_like(&mut self, value: Option<String>) -> bool {
+        match self.role_like_mut() {
+            Some(role_like) => {
+                role_like.set_reasoning_effort(value);
+                true
+            }
+            None => false,
+        }
+    }
+
     pub fn set_enabled_tools_on_role_like(&mut self, value: Option<Vec<String>>) -> bool {
         match self.role_like_mut() {
             Some(role_like) => {
@@ -1121,6 +1131,10 @@ impl RequestContext {
                 super::format_option_value(&role.temperature()),
             ),
             ("top_p", super::format_option_value(&role.top_p())),
+            (
+                "reasoning_effort",
+                super::format_option_value(&role.reasoning_effort()),
+            ),
             (
                 "enabled_tools",
                 super::format_option_value(&role.enabled_tools().map(|v| v.join(","))),
@@ -2009,6 +2023,24 @@ impl RequestContext {
                     self.update_app_config(|app| app.top_p = value);
                 }
             }
+            "reasoning_effort" => {
+                let value: Option<String> = super::parse_value(value)?;
+                if let Some(ref level) = value {
+                    let levels = self.current_model().reasoning_levels();
+                    if levels.is_empty() {
+                        bail!("The current model does not support reasoning effort configuration");
+                    }
+                    if !levels.iter().any(|l| l == level) {
+                        bail!(
+                            "Invalid reasoning effort '{level}'. Supported levels for this model: {}",
+                            levels.join(", ")
+                        );
+                    }
+                }
+                if !self.set_reasoning_effort_on_role_like(value.clone()) {
+                    self.update_app_config(|app| app.reasoning_effort = value);
+                }
+            }
             "enabled_tools" => {
                 let raw: Option<String> = super::parse_value(value)?;
                 let parsed: Option<Vec<String>> = raw.map(|s| super::csv_to_vec(&s));
@@ -2303,6 +2335,10 @@ impl RequestContext {
                     super::map_completion_values(values)
                 }
                 ".macro" => super::map_completion_values(paths::list_macros()),
+                ".reasoning" => {
+                    let levels = self.current_model().reasoning_levels();
+                    levels.iter().map(|v| (v.clone(), None)).collect()
+                }
                 ".starter" => match &self.agent {
                     Some(agent) => agent
                         .conversation_starters()
@@ -2338,6 +2374,9 @@ impl RequestContext {
                         "save",
                         "highlight",
                     ];
+                    if !self.current_model().reasoning_levels().is_empty() {
+                        values.push("reasoning_effort");
+                    }
                     values.sort_unstable();
                     values
                         .into_iter()
@@ -2507,6 +2546,10 @@ impl RequestContext {
                 }
                 "skill_instructions" => vec!["null".to_string()],
                 "memory" => super::complete_bool(self.should_inject_memory()),
+                "reasoning_effort" => {
+                    let levels = self.current_model().reasoning_levels();
+                    levels.to_vec()
+                }
                 _ => vec![],
             };
             values = candidates.into_iter().map(|v| (v, None)).collect();
@@ -2704,6 +2747,22 @@ impl RequestContext {
             None
         };
 
+        if let Some(ref effort) = role.reasoning_effort() {
+            let levels = role.model().reasoning_levels();
+            if levels.is_empty() {
+                bail!(
+                    "Role has reasoning_effort '{}' configured but the model does not support reasoning effort",
+                    effort
+                );
+            }
+            if !levels.iter().any(|l| l == effort) {
+                bail!(
+                    "Role's reasoning_effort '{}' is not valid for the model. Supported levels: {}",
+                    effort,
+                    levels.join(", ")
+                );
+            }
+        }
         self.use_role_obj(role)?;
         self.rebuild_tool_scope(app, mcp_servers, abort_signal)
             .await
@@ -2759,6 +2818,23 @@ impl RequestContext {
                 None
             };
 
+            if let Some(ref effort) = session.reasoning_effort() {
+                let levels = session.model().reasoning_levels();
+                if levels.is_empty() {
+                    bail!(
+                        "Session has reasoning_effort '{}' configured but the model does not support reasoning effort",
+                        effort
+                    );
+                }
+                if !levels.iter().any(|l| l == effort) {
+                    bail!(
+                        "Session's reasoning_effort '{}' is not valid for the model. Supported levels: {}",
+                        effort,
+                        levels.join(", ")
+                    );
+                }
+            }
+
             self.rebuild_tool_scope(app, mcp_servers, abort_signal.clone())
                 .await?;
 
@@ -2813,6 +2889,23 @@ impl RequestContext {
         )
         .await?;
 
+        if let Some(ref effort) = agent.reasoning_effort() {
+            let levels = agent.model().reasoning_levels();
+            if levels.is_empty() {
+                bail!(
+                    "Agent has reasoning_effort '{}' configured but the model does not support reasoning effort",
+                    effort
+                );
+            }
+            if !levels.iter().any(|l| l == effort) {
+                bail!(
+                    "Agent's reasoning_effort '{}' is not valid for the model. Supported levels: {}",
+                    effort,
+                    levels.join(", ")
+                );
+            }
+        }
+
         let is_graph_agent = graph::agent_has_graph(agent_name);
         if is_graph_agent && session_name.is_some() {
             bail!(

src/config/role.rs

@@ -32,7 +32,9 @@ pub trait RoleLike {
     fn enabled_mcp_servers(&self) -> Option<Vec<String>>;
     fn set_model(&mut self, model: Model);
     fn set_temperature(&mut self, value: Option<f64>);
+    fn reasoning_effort(&self) -> Option<String>;
     fn set_top_p(&mut self, value: Option<f64>);
+    fn set_reasoning_effort(&mut self, value: Option<String>);
     fn set_enabled_tools(&mut self, value: Option<Vec<String>>);
     fn set_enabled_mcp_servers(&mut self, value: Option<Vec<String>>);
 }
@@ -51,6 +53,8 @@ pub struct Role {
     temperature: Option<f64>,
     #[serde(skip_serializing_if = "Option::is_none")]
     top_p: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    reasoning_effort: Option<String>,
     #[serde(
         default,
         skip_serializing_if = "Option::is_none",
@@ -116,6 +120,9 @@ impl Role {
                     "model" => role.model_id = value.as_str().map(|v| v.to_string()),
                     "temperature" => role.temperature = value.as_f64(),
                     "top_p" => role.top_p = value.as_f64(),
+                    "reasoning_effort" => {
+                        role.reasoning_effort = value.as_str().map(|v| v.to_string())
+                    }
                     "enabled_tools" => role.enabled_tools = parse_string_or_array(value),
                     "enabled_mcp_servers" => {
                         role.enabled_mcp_servers = parse_string_or_array(value)
@@ -170,6 +177,9 @@ impl Role {
         if let Some(top_p) = self.top_p() {
             metadata.push(format!("top_p: {top_p}"));
         }
+        if let Some(reasoning_effort) = self.reasoning_effort() {
+            metadata.push(format!("reasoning_effort: {reasoning_effort}"));
+        }
         if let Some(enabled_tools) = &self.enabled_tools {
             let inline = serde_json::to_string(enabled_tools).unwrap_or_else(|_| "[]".to_string());
             metadata.push(format!("enabled_tools: {inline}"));
@@ -256,6 +266,9 @@ impl Role {
             enabled_tools,
             enabled_mcp_servers,
         );
+        if let Some(v) = role_like.reasoning_effort() {
+            self.set_reasoning_effort(Some(v));
+        }
     }
 
     pub fn batch_set(
@@ -410,6 +423,10 @@ impl RoleLike for Role {
         self.top_p
     }
 
+    fn reasoning_effort(&self) -> Option<String> {
+        self.reasoning_effort.clone()
+    }
+
     fn enabled_tools(&self) -> Option<Vec<String>> {
         self.enabled_tools.clone()
     }
@@ -433,6 +450,10 @@ impl RoleLike for Role {
         self.top_p = value;
     }
 
+    fn set_reasoning_effort(&mut self, value: Option<String>) {
+        self.reasoning_effort = value;
+    }
+
     fn set_enabled_tools(&mut self, value: Option<Vec<String>>) {
         self.enabled_tools = value;
     }

src/config/session.rs

@@ -24,6 +24,8 @@ pub struct Session {
     temperature: Option<f64>,
     #[serde(skip_serializing_if = "Option::is_none")]
     top_p: Option<f64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    reasoning_effort: Option<String>,
     #[serde(
         default,
         skip_serializing_if = "Option::is_none",
@@ -401,6 +403,7 @@ impl Session {
         self.model_id = role.model().id();
         self.temperature = role.temperature();
         self.top_p = role.top_p();
+        self.reasoning_effort = role.reasoning_effort();
         self.enabled_tools = role.enabled_tools();
         self.enabled_mcp_servers = role.enabled_mcp_servers();
         self.model = role.model().clone();
@@ -792,6 +795,10 @@ impl RoleLike for Session {
         self.top_p
     }
 
+    fn reasoning_effort(&self) -> Option<String> {
+        self.reasoning_effort.clone()
+    }
+
     fn enabled_tools(&self) -> Option<Vec<String>> {
         self.enabled_tools.clone()
     }
@@ -823,6 +830,13 @@ impl RoleLike for Session {
         }
     }
 
+    fn set_reasoning_effort(&mut self, value: Option<String>) {
+        if self.reasoning_effort != value {
+            self.reasoning_effort = value;
+            self.dirty = true;
+        }
+    }
+
     fn set_enabled_tools(&mut self, value: Option<Vec<String>>) {
         if self.enabled_tools != value {
             self.enabled_tools = value;

src/graph/llm.rs

@@ -329,6 +329,9 @@ fn build_inline_role(
     if let Some(p) = node.top_p {
         role.set_top_p(Some(p));
     }
+    if let Some(v) = &node.reasoning_effort {
+        role.set_reasoning_effort(Some(v.clone()));
+    }
 
     if node.tools.as_deref().unwrap_or_default().is_empty() {
         role.set_enabled_tools(Some(Vec::new()));
@@ -499,6 +502,7 @@ mod tests {
             model: None,
             temperature: None,
             top_p: None,
+            reasoning_effort: None,
             fallback: None,
             max_attempts: 1,
             max_iterations: 10,

src/graph/types.rs

@@ -25,6 +25,9 @@ pub struct Graph {
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub top_p: Option<f64>,
 
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_effort: Option<String>,
+
     #[serde(default)]
     pub global_tools: Vec<String>,
 
@@ -288,6 +291,9 @@ pub struct LlmNode {
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub top_p: Option<f64>,
 
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub reasoning_effort: Option<String>,
+
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub fallback: Option<String>,
 

src/graph/validator.rs

@@ -946,6 +946,7 @@ mod tests {
             model: None,
             temperature: None,
             top_p: None,
+            reasoning_effort: None,
             global_tools: Vec::new(),
             mcp_servers: Vec::new(),
             skills_enabled: None,
@@ -1048,6 +1049,7 @@ mod tests {
                 model: None,
                 temperature: None,
                 top_p: None,
+                reasoning_effort: None,
                 fallback: fallback.map(String::from),
                 max_attempts: 1,
                 max_iterations: 10,

src/rag/graph.rs

@@ -362,6 +362,7 @@ pub async fn extract_entities(
         messages,
         temperature: Some(0.0),
         top_p: None,
+        reasoning_effort: None,
         functions: None,
         stream: false,
     };

src/repl/completer.rs

@@ -31,6 +31,7 @@ impl Completer for ReplCompleter {
 
         let ctx = self.ctx.read();
         let state = ctx.state();
+        let model_has_reasoning = !ctx.current_model().reasoning_levels().is_empty();
 
         let command_filter = parts
             .iter()
@@ -44,6 +45,7 @@ impl Completer for ReplCompleter {
             .filter(|cmd| {
                 cmd.is_valid(state)
                     && (command_filter.len() == 1 || cmd.name.starts_with(&command_filter[..2]))
+                    && (cmd.name != ".reasoning" || model_has_reasoning)
             })
             .collect();
         let commands = fuzzy_filter(commands, |v| v.name, &command_filter);

src/repl/mod.rs

@@ -52,7 +52,7 @@ pub const DEFAULT_CONTINUATION_PROMPT: &str = indoc! {"
     4. Continue with the next pending item now. Call tools immediately."
 };
 
-static REPL_COMMANDS: LazyLock<[ReplCommand; 51]> = LazyLock::new(|| {
+static REPL_COMMANDS: LazyLock<[ReplCommand; 52]> = LazyLock::new(|| {
     [
         ReplCommand::new(".help", "Show this help guide", AssertState::pass()),
         ReplCommand::new(".info", "Show system info", AssertState::pass()),
@@ -259,6 +259,11 @@ static REPL_COMMANDS: LazyLock<[ReplCommand; 51]> = LazyLock::new(|| {
         ),
         ReplCommand::new(".copy", "Copy last response", AssertState::pass()),
         ReplCommand::new(".set", "Modify runtime settings", AssertState::pass()),
+        ReplCommand::new(
+            ".reasoning",
+            "Set the reasoning effort level for the current model",
+            AssertState::pass(),
+        ),
         ReplCommand::new(
             ".delete",
             "Delete roles, sessions, RAGs, or agents",
@@ -1070,6 +1075,15 @@ pub async fn run_repl_command(
                     println!("Usage: .set <key> <value>...")
                 }
             },
+            ".reasoning" => match args {
+                Some(level) => {
+                    let set_args = format!("reasoning_effort {level}");
+                    ctx.update(&set_args, abort_signal).await?;
+                }
+                None => {
+                    println!("Usage: .reasoning <level>")
+                }
+            },
             ".delete" => match args {
                 Some(args) => {
                     ctx.delete(args)?;
@@ -1600,8 +1614,8 @@ mod tests {
     }
 
     #[test]
-    fn repl_commands_has_50_entries() {
-        assert_eq!(REPL_COMMANDS.len(), 50);
+    fn repl_commands_has_52_entries() {
+        assert_eq!(REPL_COMMANDS.len(), 52);
     }
 
     #[test]