feat(platform): update OpenAI calls to use responses.create for reasoning models

Adds conditional support for OpenAI's Responses API for reasoning models
(o1, o3, etc.) that are incompatible with chat.completions.create.

Changes:
- Add openai_responses.py helper module with:
  - requires_responses_api() for model detection (exact matching)
  - convert_tools_to_responses_format() for tool format conversion
  - extract_responses_tool_calls() for tool call extraction
  - extract_usage() for normalized token usage
  - extract_responses_content() for content extraction
  - extract_responses_reasoning() for reasoning extraction
- Update llm.py OpenAI provider to conditionally use responses.create
  for reasoning models while keeping chat.completions.create for others
- Add unit tests for helper functions

Resolves: #11624
Linear: OPEN-2911

autogpt_platform/backend/backend/blocks/llm.py

@@ -32,6 +32,14 @@ from backend.data.model import (
 from backend.integrations.providers import ProviderName
 from backend.util import json
 from backend.util.logging import TruncatedLogger
+from backend.util.openai_responses import (
+    convert_tools_to_responses_format,
+    extract_responses_content,
+    extract_responses_reasoning,
+    extract_responses_tool_calls,
+    extract_usage,
+    requires_responses_api,
+)
 from backend.util.prompt import compress_context, estimate_token_count
 from backend.util.text import TextFormatter
 
@@ -659,38 +667,72 @@ async def llm_call(
     max_tokens = max(min(available_tokens, model_max_output, user_max), 1)
 
     if provider == "openai":
-        tools_param = tools if tools else openai.NOT_GIVEN
         oai_client = openai.AsyncOpenAI(api_key=credentials.api_key.get_secret_value())
-        response_format = None
 
-        parallel_tool_calls = get_parallel_tool_calls_param(
-            llm_model, parallel_tool_calls
-        )
+        # Check if this model requires the Responses API (reasoning models: o1, o3, etc.)
+        if requires_responses_api(llm_model.value):
+            # Use responses.create for reasoning models
+            tools_converted = (
+                convert_tools_to_responses_format(tools) if tools else None
+            )
 
-        if force_json_output:
-            response_format = {"type": "json_object"}
+            response = await oai_client.responses.create(
+                model=llm_model.value,
+                input=prompt,  # type: ignore
+                tools=tools_converted,  # type: ignore
+                max_output_tokens=max_tokens,
+                store=False,  # Don't persist conversations
+            )
 
-        response = await oai_client.chat.completions.create(
-            model=llm_model.value,
-            messages=prompt,  # type: ignore
-            response_format=response_format,  # type: ignore
-            max_completion_tokens=max_tokens,
-            tools=tools_param,  # type: ignore
-            parallel_tool_calls=parallel_tool_calls,
-        )
+            tool_calls = extract_responses_tool_calls(response)
+            reasoning = extract_responses_reasoning(response)
+            content = extract_responses_content(response)
+            prompt_tokens, completion_tokens = extract_usage(response, True)
 
-        tool_calls = extract_openai_tool_calls(response)
-        reasoning = extract_openai_reasoning(response)
+            return LLMResponse(
+                raw_response=response,
+                prompt=prompt,
+                response=content,
+                tool_calls=tool_calls,
+                prompt_tokens=prompt_tokens,
+                completion_tokens=completion_tokens,
+                reasoning=reasoning,
+            )
+        else:
+            # Use chat.completions.create for standard models
+            tools_param = tools if tools else openai.NOT_GIVEN
+            response_format = None
 
-        return LLMResponse(
-            raw_response=response.choices[0].message,
-            prompt=prompt,
-            response=response.choices[0].message.content or "",
-            tool_calls=tool_calls,
-            prompt_tokens=response.usage.prompt_tokens if response.usage else 0,
-            completion_tokens=response.usage.completion_tokens if response.usage else 0,
-            reasoning=reasoning,
-        )
+            parallel_tool_calls = get_parallel_tool_calls_param(
+                llm_model, parallel_tool_calls
+            )
+
+            if force_json_output:
+                response_format = {"type": "json_object"}
+
+            response = await oai_client.chat.completions.create(
+                model=llm_model.value,
+                messages=prompt,  # type: ignore
+                response_format=response_format,  # type: ignore
+                max_completion_tokens=max_tokens,
+                tools=tools_param,  # type: ignore
+                parallel_tool_calls=parallel_tool_calls,
+            )
+
+            tool_calls = extract_openai_tool_calls(response)
+            reasoning = extract_openai_reasoning(response)
+
+            return LLMResponse(
+                raw_response=response.choices[0].message,
+                prompt=prompt,
+                response=response.choices[0].message.content or "",
+                tool_calls=tool_calls,
+                prompt_tokens=response.usage.prompt_tokens if response.usage else 0,
+                completion_tokens=(
+                    response.usage.completion_tokens if response.usage else 0
+                ),
+                reasoning=reasoning,
+            )
     elif provider == "anthropic":
 
         an_tools = convert_openai_tool_fmt_to_anthropic(tools)

autogpt_platform/backend/backend/util/openai_responses.py

@@ -0,0 +1,185 @@
+"""Helpers for OpenAI Responses API migration.
+
+This module provides utilities for conditionally using OpenAI's Responses API
+instead of Chat Completions for reasoning models (o1, o3, etc.) that require it.
+"""
+
+from typing import Any
+
+# Exact model identifiers that require the Responses API.
+# Use exact matching to avoid false positives on future models.
+# NOTE: Update this set when OpenAI releases new reasoning models.
+REASONING_MODELS = frozenset(
+    {
+        # O1 family
+        "o1",
+        "o1-mini",
+        "o1-preview",
+        "o1-2024-12-17",
+        # O3 family
+        "o3",
+        "o3-mini",
+        "o3-2025-04-16",
+        "o3-mini-2025-01-31",
+    }
+)
+
+
+def requires_responses_api(model: str) -> bool:
+    """Check if model requires the Responses API (exact match).
+
+    Args:
+        model: The model identifier string (e.g., "o3-mini", "gpt-4o")
+
+    Returns:
+        True if the model requires responses.create, False otherwise
+    """
+    return model in REASONING_MODELS
+
+
+def convert_tools_to_responses_format(tools: list[dict] | None) -> list[dict]:
+    """Convert Chat Completions tool format to Responses API format.
+
+    The Responses API uses internally-tagged polymorphism (flatter structure)
+    and functions are strict by default.
+
+    Chat Completions format:
+        {"type": "function", "function": {"name": "...", "parameters": {...}}}
+
+    Responses API format:
+        {"type": "function", "name": "...", "parameters": {...}}
+
+    Args:
+        tools: List of tools in Chat Completions format
+
+    Returns:
+        List of tools in Responses API format
+    """
+    if not tools:
+        return []
+
+    converted = []
+    for tool in tools:
+        if tool.get("type") == "function":
+            func = tool.get("function", {})
+            converted.append(
+                {
+                    "type": "function",
+                    "name": func.get("name"),
+                    "description": func.get("description"),
+                    "parameters": func.get("parameters"),
+                    # Note: strict=True is default in Responses API
+                }
+            )
+        else:
+            # Pass through non-function tools as-is
+            converted.append(tool)
+    return converted
+
+
+def extract_responses_tool_calls(response: Any) -> list[dict] | None:
+    """Extract tool calls from Responses API response.
+
+    The Responses API returns tool calls as separate items in the output array
+    with type="function_call".
+
+    Args:
+        response: The Responses API response object
+
+    Returns:
+        List of tool calls in a normalized format, or None if no tool calls
+    """
+    tool_calls = []
+    for item in response.output:
+        if getattr(item, "type", None) == "function_call":
+            tool_calls.append(
+                {
+                    "id": item.call_id,
+                    "type": "function",
+                    "function": {
+                        "name": item.name,
+                        "arguments": item.arguments,
+                    },
+                }
+            )
+    return tool_calls if tool_calls else None
+
+
+def extract_usage(response: Any, is_responses_api: bool) -> tuple[int, int]:
+    """Extract token usage from either API response.
+
+    The Responses API uses different field names for token counts:
+    - Chat Completions: prompt_tokens, completion_tokens
+    - Responses API: input_tokens, output_tokens
+
+    Args:
+        response: The API response object
+        is_responses_api: True if response is from Responses API
+
+    Returns:
+        Tuple of (prompt_tokens, completion_tokens)
+    """
+    if not response.usage:
+        return 0, 0
+
+    if is_responses_api:
+        # Responses API uses different field names
+        return (
+            getattr(response.usage, "input_tokens", 0),
+            getattr(response.usage, "output_tokens", 0),
+        )
+    else:
+        # Chat Completions API
+        return (
+            getattr(response.usage, "prompt_tokens", 0),
+            getattr(response.usage, "completion_tokens", 0),
+        )
+
+
+def extract_responses_content(response: Any) -> str:
+    """Extract text content from Responses API response.
+
+    Args:
+        response: The Responses API response object
+
+    Returns:
+        The text content from the response, or empty string if none
+    """
+    # The SDK provides a helper property
+    if hasattr(response, "output_text"):
+        return response.output_text or ""
+
+    # Fallback: manually extract from output items
+    for item in response.output:
+        if getattr(item, "type", None) == "message":
+            for content in getattr(item, "content", []):
+                if getattr(content, "type", None) == "output_text":
+                    return getattr(content, "text", "")
+    return ""
+
+
+def extract_responses_reasoning(response: Any) -> str | None:
+    """Extract reasoning content from Responses API response.
+
+    Reasoning models return their reasoning process in the response,
+    which can be useful for debugging or display.
+
+    Args:
+        response: The Responses API response object
+
+    Returns:
+        The reasoning text, or None if not present
+    """
+    for item in response.output:
+        if getattr(item, "type", None) == "reasoning":
+            # Reasoning items may have summary or content
+            summary = getattr(item, "summary", [])
+            if summary:
+                # Join summary items if present
+                texts = []
+                for s in summary:
+                    if hasattr(s, "text"):
+                        texts.append(s.text)
+                if texts:
+                    return "\n".join(texts)
+    return None

autogpt_platform/backend/backend/util/openai_responses_test.py

@@ -0,0 +1,155 @@
+"""Tests for OpenAI Responses API helpers."""
+
+import pytest
+
+from backend.util.openai_responses import (
+    REASONING_MODELS,
+    convert_tools_to_responses_format,
+    requires_responses_api,
+)
+
+
+class TestRequiresResponsesApi:
+    """Tests for the requires_responses_api function."""
+
+    def test_o1_models_require_responses_api(self):
+        """O1 family models should require the Responses API."""
+        assert requires_responses_api("o1") is True
+        assert requires_responses_api("o1-mini") is True
+        assert requires_responses_api("o1-preview") is True
+        assert requires_responses_api("o1-2024-12-17") is True
+
+    def test_o3_models_require_responses_api(self):
+        """O3 family models should require the Responses API."""
+        assert requires_responses_api("o3") is True
+        assert requires_responses_api("o3-mini") is True
+        assert requires_responses_api("o3-2025-04-16") is True
+        assert requires_responses_api("o3-mini-2025-01-31") is True
+
+    def test_gpt_models_do_not_require_responses_api(self):
+        """GPT models should NOT require the Responses API."""
+        assert requires_responses_api("gpt-4o") is False
+        assert requires_responses_api("gpt-4o-mini") is False
+        assert requires_responses_api("gpt-4-turbo") is False
+        assert requires_responses_api("gpt-3.5-turbo") is False
+        assert requires_responses_api("gpt-5") is False
+        assert requires_responses_api("gpt-5-mini") is False
+
+    def test_other_models_do_not_require_responses_api(self):
+        """Other provider models should NOT require the Responses API."""
+        assert requires_responses_api("claude-3-opus") is False
+        assert requires_responses_api("llama-3.3-70b") is False
+        assert requires_responses_api("gemini-pro") is False
+
+    def test_empty_string_does_not_require_responses_api(self):
+        """Empty string should not require the Responses API."""
+        assert requires_responses_api("") is False
+
+    def test_exact_matching_no_false_positives(self):
+        """Should not match models that just start with 'o1' or 'o3'."""
+        # These are hypothetical models that start with o1/o3 but aren't
+        # actually reasoning models
+        assert requires_responses_api("o1-turbo-hypothetical") is False
+        assert requires_responses_api("o3-fast-hypothetical") is False
+        assert requires_responses_api("o100") is False
+
+
+class TestConvertToolsToResponsesFormat:
+    """Tests for the convert_tools_to_responses_format function."""
+
+    def test_empty_tools_returns_empty_list(self):
+        """Empty or None tools should return empty list."""
+        assert convert_tools_to_responses_format(None) == []
+        assert convert_tools_to_responses_format([]) == []
+
+    def test_converts_function_tool_format(self):
+        """Should convert Chat Completions function format to Responses format."""
+        chat_completions_tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "get_weather",
+                    "description": "Get the weather in a location",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {
+                            "location": {"type": "string"},
+                        },
+                        "required": ["location"],
+                    },
+                },
+            }
+        ]
+
+        result = convert_tools_to_responses_format(chat_completions_tools)
+
+        assert len(result) == 1
+        assert result[0]["type"] == "function"
+        assert result[0]["name"] == "get_weather"
+        assert result[0]["description"] == "Get the weather in a location"
+        assert result[0]["parameters"] == {
+            "type": "object",
+            "properties": {
+                "location": {"type": "string"},
+            },
+            "required": ["location"],
+        }
+        # Should not have nested "function" key
+        assert "function" not in result[0]
+
+    def test_handles_multiple_tools(self):
+        """Should handle multiple tools."""
+        chat_completions_tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "tool_1",
+                    "description": "First tool",
+                    "parameters": {"type": "object", "properties": {}},
+                },
+            },
+            {
+                "type": "function",
+                "function": {
+                    "name": "tool_2",
+                    "description": "Second tool",
+                    "parameters": {"type": "object", "properties": {}},
+                },
+            },
+        ]
+
+        result = convert_tools_to_responses_format(chat_completions_tools)
+
+        assert len(result) == 2
+        assert result[0]["name"] == "tool_1"
+        assert result[1]["name"] == "tool_2"
+
+    def test_passes_through_non_function_tools(self):
+        """Non-function tools should be passed through as-is."""
+        tools = [{"type": "web_search", "config": {"enabled": True}}]
+
+        result = convert_tools_to_responses_format(tools)
+
+        assert result == tools
+
+
+class TestReasoningModelsSet:
+    """Tests for the REASONING_MODELS constant."""
+
+    def test_reasoning_models_is_frozenset(self):
+        """REASONING_MODELS should be a frozenset (immutable)."""
+        assert isinstance(REASONING_MODELS, frozenset)
+
+    def test_contains_expected_models(self):
+        """Should contain all expected reasoning models."""
+        expected = {
+            "o1",
+            "o1-mini",
+            "o1-preview",
+            "o1-2024-12-17",
+            "o3",
+            "o3-mini",
+            "o3-2025-04-16",
+            "o3-mini-2025-01-31",
+        }
+        assert expected.issubset(REASONING_MODELS)

autogpt_platform/frontend/instrumentation-client.ts

@@ -22,6 +22,11 @@ Sentry.init({
 
   enabled: shouldEnable,
 
+  // Suppress cross-origin stylesheet errors from Sentry Replay (rrweb)
+  // serializing DOM snapshots with cross-origin stylesheets
+  // (e.g., from browser extensions or CDN-loaded CSS)
+  ignoreErrors: [/Not allowed to access cross-origin stylesheet/],
+
   // Add optional integrations for additional features
   integrations: [
     Sentry.captureConsoleIntegration(),

autogpt_platform/frontend/src/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/sidebar/SidebarRunsList/components/ScheduleListItem.tsx

@@ -29,6 +29,7 @@ export function ScheduleListItem({
       description={formatDistanceToNow(schedule.next_run_time, {
         addSuffix: true,
       })}
+      descriptionTitle={new Date(schedule.next_run_time).toString()}
       onClick={onClick}
       selected={selected}
       icon={

autogpt_platform/frontend/src/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/sidebar/SidebarRunsList/components/SidebarItemCard.tsx

@@ -7,6 +7,7 @@ import React from "react";
 interface Props {
   title: string;
   description?: string;
+  descriptionTitle?: string;
   icon?: React.ReactNode;
   selected?: boolean;
   onClick?: () => void;
@@ -16,6 +17,7 @@ interface Props {
 export function SidebarItemCard({
   title,
   description,
+  descriptionTitle,
   icon,
   selected,
   onClick,
@@ -38,7 +40,11 @@ export function SidebarItemCard({
           >
             {title}
           </Text>
-          <Text variant="body" className="leading-tight !text-zinc-500">
+          <Text
+            variant="body"
+            className="leading-tight !text-zinc-500"
+            title={descriptionTitle}
+          >
             {description}
           </Text>
         </div>

autogpt_platform/frontend/src/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/sidebar/SidebarRunsList/components/TaskListItem.tsx

@@ -81,6 +81,9 @@ export function TaskListItem({
           ? formatDistanceToNow(run.started_at, { addSuffix: true })
           : "—"
       }
+      descriptionTitle={
+        run.started_at ? new Date(run.started_at).toString() : undefined
+      }
       onClick={onClick}
       selected={selected}
       actions={

autogpt_platform/frontend/src/app/globals.css

@@ -180,3 +180,14 @@ body[data-google-picker-open="true"] [data-dialog-content] {
   z-index: 1 !important;
   pointer-events: none !important;
 }
+
+/* CoPilot chat table styling — remove left/right borders, increase padding */
+[data-streamdown="table-wrapper"] table {
+  border-left: none;
+  border-right: none;
+}
+
+[data-streamdown="table-wrapper"] th,
+[data-streamdown="table-wrapper"] td {
+  padding: 0.875rem 1rem; /* py-3.5 px-4 */
+}

autogpt_platform/frontend/src/components/contextual/OutputRenderers/renderers/MarkdownRenderer.tsx

@@ -226,7 +226,7 @@ function renderMarkdown(
           table: ({ children, ...props }) => (
             <div className="my-4 overflow-x-auto">
               <table
-                className="min-w-full divide-y divide-gray-200 rounded-lg border border-gray-200 dark:divide-gray-700 dark:border-gray-700"
+                className="min-w-full divide-y divide-gray-200 border-y border-gray-200 dark:divide-gray-700 dark:border-gray-700"
                 {...props}
               >
                 {children}
@@ -235,7 +235,7 @@ function renderMarkdown(
           ),
           th: ({ children, ...props }) => (
             <th
-              className="bg-gray-50 px-4 py-3 text-left text-xs font-semibold uppercase tracking-wider text-gray-700 dark:bg-gray-800 dark:text-gray-300"
+              className="bg-gray-50 px-4 py-3.5 text-left text-xs font-semibold uppercase tracking-wider text-gray-700 dark:bg-gray-800 dark:text-gray-300"
               {...props}
             >
               {children}
@@ -243,7 +243,7 @@ function renderMarkdown(
           ),
           td: ({ children, ...props }) => (
             <td
-              className="border-t border-gray-200 px-4 py-3 text-sm text-gray-600 dark:border-gray-700 dark:text-gray-400"
+              className="border-t border-gray-200 px-4 py-3.5 text-sm text-gray-600 dark:border-gray-700 dark:text-gray-400"
               {...props}
             >
               {children}