Thinking

Thinking (or reasoning) is the process by which a model works through a problem step-by-step before providing its final answer.

The simplest way to enable thinking across supported providers is the Thinking capability. Provider-specific settings are available for advanced usage when you need direct access to a provider’s native thinking controls.

Unified thinking settings

Use the Thinking capability to enable thinking:

thinking_capability.py

from pydantic_ai import Agent
from pydantic_ai.capabilities import Thinking

agent = Agent('anthropic:claude-opus-4-7', capabilities=[Thinking(effort='high')])

You can also set the underlying thinking field in ModelSettings directly:

unified_thinking.py

from pydantic_ai import Agent

agent = Agent('anthropic:claude-opus-4-7', model_settings={'thinking': 'high'})

The Thinking.effort value accepts:

• True — enable thinking with the provider’s default effort level
• False — disable thinking (silently ignored on always-on models)
• 'minimal' / 'low' / 'medium' / 'high' / 'xhigh' — enable thinking at a specific effort level (unsupported levels map to the closest available value)

These are the same values accepted by the underlying thinking model setting. When omitted, the model uses its default behavior. Provider-specific settings (documented in the sections below) take precedence when both are set.

Provider translation

The Thinking capability maps each effort value to the selected provider’s native format:

Provider	Thinking() / Thinking(effort=True)	Thinking(effort='high')	Notes
Anthropic (Opus 4.6+)	anthropic_thinking={'type': 'adaptive'}	{type: 'adaptive'} + effort='high'	Claude Opus 4.7 and 4.8 also support effort='xhigh'
Anthropic (older)	anthropic_thinking={'type': 'enabled', 'budget_tokens': 10000}	budget_tokens=16384	Budget-based; 'low' → 2048 tokens
OpenAI	reasoning_effort='medium'	reasoning_effort='high'
Google (Gemini 3+)	include_thoughts=True	thinking_level='HIGH'
Google (Gemini 2.5)	include_thoughts=True	thinking_budget=24576
Groq	reasoning_format='parsed'	reasoning_format='parsed'	thinking=False → 'hidden' (no true disable)
OpenRouter	reasoning={'effort': 'medium', 'enabled': True}	reasoning={'effort': 'high', 'enabled': True}	thinking=False → effort='none'; always-on routes silently ignore; via extra_body
Cerebras	disable_reasoning=False	disable_reasoning=False	thinking=False → disable_reasoning=True
xAI	reasoning_effort omitted on Grok 4.3 (uses its default)	reasoning_effort='high'	Grok 4.3 supports 'none', 'low', 'medium', and 'high', and thinking=True omits the parameter so the model applies its own default; Grok 3 Mini only supports 'low' and 'high' (so thinking=True → 'high') and silently ignores thinking=False
Bedrock (Claude 4.6+)	thinking.type='adaptive'	{type: 'adaptive'} + output_config.effort='high'	Effort lives in the sibling output_config field per AWS docs; xhigh maps to max
Bedrock (Claude older)	thinking.type='enabled'	budget_tokens=16384	Budget-based
Bedrock (OpenAI)	reasoning_effort='medium'	reasoning_effort='high'	Converse rejects 'none'; thinking=False silently ignored
Bedrock (Qwen)	reasoning_config='high'	reasoning_config='high'	Only 'low' and 'high'; thinking=False silently ignored

OpenAI

When using the OpenAIChatModel, text output inside <think> tags are converted to ThinkingPart objects. You can customize the tags using the thinking_tags field on the model profile.

Some OpenAI-compatible model providers might also support native thinking parts that are not delimited by tags. Instead, they are sent and received as separate, custom fields in the API. Typically, if you are calling the model via the <provider>:<model> shorthand, Pydantic AI handles it for you. Nonetheless, you can still configure the fields with openai_chat_thinking_field.

If your provider recommends to send back these custom fields not changed, for caching or interleaved thinking benefits, you can also achieve this with openai_chat_send_back_thinking_parts.

OpenAI Responses

The OpenAIResponsesModel can generate native thinking parts. To enable this functionality, you need to set the OpenAIResponsesModelSettings.openai_reasoning_effort and OpenAIResponsesModelSettings.openai_reasoning_summary model settings.

By default, the unique IDs of reasoning, text, and function call parts from the message history are sent to the model, which can result in errors like "Item 'rs_123' of type 'reasoning' was provided without its required following item." if the message history you’re sending does not match exactly what was received from the Responses API in a previous response, for example if you’re using a history processor. To disable this, you can disable the OpenAIResponsesModelSettings.openai_send_reasoning_ids model setting.

openai_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIResponsesModel, OpenAIResponsesModelSettings

model = OpenAIResponsesModel('gpt-5.2')
settings = OpenAIResponsesModelSettings(
    openai_reasoning_effort='low',
    openai_reasoning_summary='detailed',
)
agent = Agent(model, model_settings=settings)
...

Raw reasoning without summaries

Some OpenAI-compatible APIs (such as LM Studio, vLLM, or OpenRouter with gpt-oss models) may return raw reasoning content without reasoning summaries. In this case, ThinkingPart.content will be empty, but the raw reasoning is available in provider_details['raw_content']. Following OpenAI’s guidance that raw reasoning should not be shown directly to users, we store it in provider_details rather than in the main content field.

Anthropic

To enable thinking, use the AnthropicModelSettings.anthropic_thinking model setting.

Note

Extended thinking (type: 'enabled' with budget_tokens) is deprecated on claude-opus-4-6 and removed on claude-opus-4-7 and claude-opus-4-8. For those models, use adaptive thinking instead.

anthropic_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.anthropic import AnthropicModel, AnthropicModelSettings

model = AnthropicModel('claude-sonnet-4-5')
settings = AnthropicModelSettings(
    anthropic_thinking={'type': 'enabled', 'budget_tokens': 1024},
)
agent = Agent(model, model_settings=settings)
...

Interleaved Thinking

To enable interleaved thinking, you need to include the beta header in your model settings:

anthropic_interleaved_thinking.py

from pydantic_ai import Agent
from pydantic_ai.models.anthropic import AnthropicModel, AnthropicModelSettings

model = AnthropicModel('claude-sonnet-4-5')
settings = AnthropicModelSettings(
    anthropic_thinking={'type': 'enabled', 'budget_tokens': 10000},
    extra_headers={'anthropic-beta': 'interleaved-thinking-2025-05-14'},
)
agent = Agent(model, model_settings=settings)
...

Adaptive Thinking & Effort

Starting with claude-opus-4-6, Anthropic supports adaptive thinking, where the model dynamically decides when and how much to think based on the complexity of each request. This replaces extended thinking (type: 'enabled' with budget_tokens) which is deprecated on Opus 4.6 and removed on Opus 4.7 and 4.8. Claude Opus 4.7 and 4.8 also add the xhigh effort level. Adaptive thinking also automatically enables interleaved thinking.

anthropic_adaptive_thinking.py

from pydantic_ai import Agent
from pydantic_ai.models.anthropic import AnthropicModel, AnthropicModelSettings

model = AnthropicModel('claude-opus-4-8')
settings = AnthropicModelSettings(
    anthropic_thinking={'type': 'adaptive'},
    anthropic_effort='high',
)
agent = Agent(model, model_settings=settings)
...

The anthropic_effort setting controls how much effort the model puts into its response (independent of thinking). See the Anthropic effort docs for details.

Note

Older models (claude-sonnet-4-5, claude-opus-4-5, etc.) do not support adaptive thinking and require {'type': 'enabled', 'budget_tokens': N} as shown above.

Thinking tokens count against Anthropic’s loop-wide task budgets, so adaptive thinking naturally scales down as the budget depletes.

Google

For advanced usage, use the GoogleModelSettings.google_thinking_config model setting.

google_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.google import GoogleModel, GoogleModelSettings

model = GoogleModel('gemini-3.5-flash')
settings = GoogleModelSettings(google_thinking_config={'include_thoughts': True, 'thinking_level': 'MEDIUM'})
agent = Agent(model, model_settings=settings)
...

See the Google model docs for more details.

xAI

xAI reasoning models (Grok) support native thinking. To preserve the thinking content for multi-turn conversations, enable XaiModelSettings.xai_include_encrypted_content.

xai_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.xai import XaiModel, XaiModelSettings

model = XaiModel('grok-4.3')
settings = XaiModelSettings(xai_include_encrypted_content=True)
agent = Agent(model, model_settings=settings)
...

Bedrock

For Claude Sonnet 4.6+ and Opus 4.6+, Pydantic AI’s unified thinking setting translates to AWS’s required adaptive thinking shape automatically — set ModelSettings.thinking and you’re done.

For older Claude models or to pin a specific budget_tokens, you can still use BedrockModelSettings.bedrock_additional_model_requests_fields model setting to pass provider-specific configuration directly:

Claude

bedrock_claude_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.bedrock import BedrockConverseModel, BedrockModelSettings

model = BedrockConverseModel('us.anthropic.claude-sonnet-4-5-20250929-v1:0')
model_settings = BedrockModelSettings(
    bedrock_additional_model_requests_fields={
        'thinking': {'type': 'enabled', 'budget_tokens': 1024}
    }
)
agent = Agent(model=model, model_settings=model_settings)

OpenAI

bedrock_openai_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.bedrock import BedrockConverseModel, BedrockModelSettings

model = BedrockConverseModel('openai.gpt-oss-120b-1:0')
model_settings = BedrockModelSettings(
    bedrock_additional_model_requests_fields={'reasoning_effort': 'low'}
)
agent = Agent(model=model, model_settings=model_settings)

Qwen

bedrock_qwen_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.bedrock import BedrockConverseModel, BedrockModelSettings

model = BedrockConverseModel('qwen.qwen3-32b-v1:0')
model_settings = BedrockModelSettings(
    bedrock_additional_model_requests_fields={'reasoning_config': 'high'}
)
agent = Agent(model=model, model_settings=model_settings)

Deepseek

Reasoning is always enabled for Deepseek model

bedrock_deepseek_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.bedrock import BedrockConverseModel

model = BedrockConverseModel('us.deepseek.r1-v1:0')
agent = Agent(model=model)

Groq

Groq supports different formats to receive thinking parts:

• "raw": The thinking part is included in the text content inside <think> tags, which are automatically converted to ThinkingPart objects.
• "hidden": The thinking part is not included in the text content.
• "parsed": The thinking part has its own structured part in the response which is converted into a ThinkingPart object.

To enable thinking, use the GroqModelSettings.groq_reasoning_format model setting:

groq_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.groq import GroqModel, GroqModelSettings

model = GroqModel('qwen/qwen3-32b')
settings = GroqModelSettings(groq_reasoning_format='parsed')
agent = Agent(model, model_settings=settings)
...

Note

Groq does not support truly disabling thinking. When thinking=False is set via the unified setting, Pydantic AI sends reasoning_format='hidden', which suppresses reasoning output but the model may still reason internally.

OpenRouter

To enable thinking, use the OpenRouterModelSettings.openrouter_reasoning model setting.

openrouter_thinking_part.py

from pydantic_ai import Agent
from pydantic_ai.models.openrouter import OpenRouterModel, OpenRouterModelSettings

model = OpenRouterModel('openai/gpt-5.2')
settings = OpenRouterModelSettings(openrouter_reasoning={'effort': 'high'})
agent = Agent(model, model_settings=settings)
...

Wire format details

Truthy thinking values send both effort and enabled: True on the wire. The explicit enabled: True is a no-op for reasoning-by-default models but load-bearing for reasoning-optional routes (parts of the google/gemma-* family, for example) that otherwise leave reasoning disabled despite effort being set.

thinking=False sends reasoning={'effort': 'none'} — the documented OpenRouter disable signal — on routes whose upstream can honor disable (e.g. anthropic/claude-sonnet-4.5, z-ai/glm-4.6). On routes whose upstream is always-on (e.g. openai/o3, openai/gpt-5, mistralai/magistral-medium-*, deepseek/deepseek-r1, x-ai/grok-3-mini), thinking=False is silently ignored at the model-profile gate, matching the same model’s direct-route behavior. Set OpenRouterModelSettings.openrouter_reasoning directly when you want explicit per-route control.

Mistral

Thinking is supported by the magistral family of models. It does not need to be specifically enabled.

Cohere

Thinking is supported by the command-a-reasoning-08-2025 model. It does not need to be specifically enabled.

Hugging Face

Text output inside <think> tags is automatically converted to ThinkingPart objects. You can customize the tags using the thinking_tags field on the model profile.

Outlines

Some local models run through Outlines include in their text output a thinking part delimited by tags. In that case, it will be handled by Pydantic AI that will separate the thinking part from the final answer without the need to specifically enable it. The thinking tags used by default are "<think>" and "</think>". If your model uses different tags, you can specify them in the model profile using the thinking_tags field.

Outlines currently does not support thinking along with structured output. If you provide an output_type, the model text output will not contain a thinking part with the associated tags, and you may experience degraded performance.