[Release] openai/openai-agents-python v0.16.0: v0.16.0

OpenAI Agents Python SDK v0.16.0 introduces significant updates including a default model change from gpt-4.1 to gpt-5.4-mini, new configuration options for disabling max_turns limits, SDK-side tool execution concurrency control, and server-prefixed MCP tool naming to prevent conflicts. The release also includes stability fixes for tool call output indexes and security improvements for symlink handling. These changes enhance performance and provide greater flexibility for agent configuration while maintaining backward compatibility through explicit model specification.

Key Points

• •Default model upgraded to gpt-5.4-mini with GPT-5 defaults (reasoning.effort='none', verbosity='low') — explicitly set model if you need gpt-4.1 behavior
• •New max_turns=None option disables run turn limits while preserving DEFAULT_MAX_TURNS (10) when omitted
• •ToolExecutionConfig(max_function_tool_concurrency=...) added to RunConfig for SDK-side local function tool execution scheduling
• •MCPConfig.include_server_in_tool_names option prevents tool name conflicts across multiple MCP servers by prefixing server names
• •Chat completions tool call output indexes stabilized for more reliable tool execution
• •External symlink targets rejected during hydrate for improved security
• •Permissions class made hashable to match User and Group behavior
• •Set OPENAI_DEFAULT_MODEL environment variable or Agent(model='gpt-4.1') to maintain previous default behavior
• •Provider-side ModelSettings.parallel_tool_calls kept separate from SDK-side local execution scheduling
• •Documentation updates for realtime guardrail fallback behavior and translated pages

Workflow Diagram

Concepts

Artifacts (4)

agent = Agent(
    name="Assistant",
    mcp_servers=[my_mcp_server],
    mcp_config={"include_server_in_tool_names": True},
)

agent = Agent(
    name="Assistant",
    model="gpt-4.1"
)

from openai.agents import RunConfig, ToolExecutionConfig

run_config = RunConfig(
    tool_execution_config=ToolExecutionConfig(
        max_function_tool_concurrency=5
    )
)

from openai.agents import RunConfig

run_config = RunConfig(
    max_turns=None  # Disable turn limit
)