pydantic-ai

v1.103.0 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 7d AI Agents & Assistants

✓ No known CVEs patched

✓ No known CVEs patched in this version

Topics

agent-framework genai llm pydantic python

Summary

AI summary

Updates 🐛 Bug Fixes, 🚀 Features, and bedrock across a mixed release.

Type	Severity	Summary	CVE
Feature
Feature	Low	Add `list_prompts` and `get_prompt` to McpServer. Add `list_prompts` and `get_prompt` to McpServer. Source: llm_adapter@2026-05-27 Confidence: high	—
Feature	Low	Round‑trip message timestamps through VercelAIAdapter's UIMessage.metadata. Round‑trip message timestamps through VercelAIAdapter's UIMessage.metadata. Source: llm_adapter@2026-05-27 Confidence: high	—
Feature	Low	Support `anthropic_eager_input_streaming` in OpenRouterModel. Support `anthropic_eager_input_streaming` in OpenRouterModel. Source: llm_adapter@2026-05-27 Confidence: high	—
Bugfix
Bugfix	Medium	Map `malformed_model_output` and `malformed_tool_use` to `FinishReason.error` in bedrock. Map `malformed_model_output` and `malformed_tool_use` to `FinishReason.error` in bedrock. Source: llm_adapter@2026-05-27 Confidence: high	—
Bugfix	Medium	Recognize `type='adaptive'` in `_is_thinking_enabled` for bedrock. Recognize `type='adaptive'` in `_is_thinking_enabled` for bedrock. Source: llm_adapter@2026-05-27 Confidence: high	—
Bugfix	Medium	Strip `force_download` flag from client‑submitted `FileUrl` parts in UIAdapter.sanitize_messages. Strip `force_download` flag from client‑submitted `FileUrl` parts in UIAdapter.sanitize_messages. Source: llm_adapter@2026-05-27 Confidence: high	—
Bugfix	Medium	Warn when prepare callbacks return `None` in toolsets. Warn when prepare callbacks return `None` in toolsets. Source: llm_adapter@2026-05-27 Confidence: high	—

Full changelog

Add list_prompts and get_prompt functionality to McpServer by @bdore in https://github.com/pydantic/pydantic-ai/pull/3889
Round-trip message timestamps through VercelAIAdapter's UIMessage.metadata by @Genmin in https://github.com/pydantic/pydantic-ai/pull/5279
Support anthropic_eager_input_streaming in OpenRouterModel by @Alex-Resch in https://github.com/pydantic/pydantic-ai/pull/5656

fix(bedrock): map malformed_model_output and malformed_tool_use to FinishReason.error by @kuishou68 in https://github.com/pydantic/pydantic-ai/pull/5640
fix(bedrock): recognize type='adaptive' in _is_thinking_enabled by @sevakva in https://github.com/pydantic/pydantic-ai/pull/5651
fix(ui): Strip force_download flag from client-submitted FileUrl parts in UIAdapter.sanitize_messages by @colesmcintosh in https://github.com/pydantic/pydantic-ai/pull/5571
fix(toolsets): warn when prepare callbacks return None by @dfm88 in https://github.com/pydantic/pydantic-ai/pull/5188

@ejentum made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5594
@kuishou68 made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5640
@sevakva made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5651
@bdore made their first contribution in https://github.com/pydantic/pydantic-ai/pull/3889
@Genmin made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5279
@aryanmotgi made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5667
@dfm88 made their first contribution in https://github.com/pydantic/pydantic-ai/pull/5188

Full Changelog: https://github.com/pydantic/pydantic-ai/compare/v1.102.0...v1.103.0

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track pydantic-ai

Get notified when new releases ship.

About pydantic-ai

AI Agent Framework, the Pydantic way