Skip to content

FastroAI Changelog

Introduction

The Changelog documents all notable changes made to FastroAI. This includes new features, bug fixes, and improvements. It's organized by version and date, providing a clear history of the library's development.

[0.5.0] - May 6, 2026

Fixed

  • FastroAgent timeout is now actually enforced by @igorbenav
  • The timeout parameter on FastroAgent was previously stored on AgentConfig but never read by _execute(). The configured value silently never reached the underlying model client, so a single LLM call could run for the full underlying httpx read timeout (~600s) × max_retries+1 attempts before any error surfaced.
  • New _build_default_model_settings() helper forwards self.config.timeout into ModelSettings.timeout, where pydantic-ai passes it through to the model client (e.g. OpenAI SDK's chat.completions.create(timeout=N)).
  • Contributed to a 50-minute production hang in a downstream worker on 2026-05-06 — a hung pydantic-ai agent call that should have hit a 300s timeout per the configured constant ran for ~30 min until the worker was manually restarted.

Changed (Breaking)

  • AgentConfig.timeout_seconds renamed to AgentConfig.timeout to match pydantic-ai's ModelSettings.timeout and to align with what FastroAgent(timeout=N) callers were already passing. Code using the old name will fail with a Pydantic validation error.
  • Default is now None (opt-in), previously 120. With None, no per-request timeout is forwarded and the model client's own default applies (typically 600s read on OpenAI). Code that wants the previous 120s behavior — though no such code worked correctly because the default was never enforced — should pass timeout=120 explicitly.
  • DEFAULT_TIMEOUT_SECONDS removed from fastroai.agent exports. No replacement constant; choose a value at the call site.

Migration

# Before:
from fastroai.agent import DEFAULT_TIMEOUT_SECONDS
agent = FastroAgent(model="openai:gpt-4o", timeout_seconds=60)
config = AgentConfig(timeout_seconds=60)

# After:
agent = FastroAgent(model="openai:gpt-4o", timeout=60)
config = AgentConfig(timeout=60)

If you have a *_TIMEOUT_SECONDS constant that you've been passing into FastroAgent and trusting to bound LLM call wall time, that bound was not actually applying. Re-validate your value — you may have been over-budgeting because timeouts were getting absorbed by retries that no longer happen as silently.

Documentation

  • Updated RELEASE.md with 0.5.0 release notes including breaking-change migration guide
  • Updated docs/guides/fastro-agent.md configuration reference table

Full Changelog: https://github.com/benavlabs/fastroai/compare/v0.4.1...v0.5.0

[0.4.1] - Jan 25, 2026

Fixed

  • FallbackModel Tracking by @igorbenav
  • FastroAgent now correctly tracks the actual model that processed the request when using PydanticAI's FallbackModel
  • Model name is extracted from ModelResponse.model_name in the message history instead of non-existent usage.model

  • Fixes incorrect cost calculations when using model wrappers (e.g., DeepSeek primary with GPT-4o fallback)

  • No False Model Assumptions by @igorbenav

  • When using escape hatch (agent=) without explicit model=, FastroAgent no longer assumes gpt-4o
  • If model can't be detected from response, returns model=None and cost_microcents=0 with a warning
  • Tokens are still tracked accurately even when model is unknown

Changed

  • ChatResponse.model is now str | None (was str) to represent unknown models
  • CostCalculator.calculate_cost() accepts model: str | None and returns 0 for None
  • Added logging when model detection fails and no fallback is configured

Documentation

  • Updated RELEASE.md with 0.4.1 release notes

Full Changelog: https://github.com/benavlabs/fastroai/compare/v0.4.0...v0.4.1

[0.4.0] - Dec 20, 2025

Added

  • Enhanced Cost Tracking by @igorbenav
  • Cache token tracking (cache_read_tokens, cache_write_tokens) for accurate cost calculation with prompt caching
  • Audio token tracking for multimodal models
  • Request count tracking (request_count) for API call monitoring
  • Tool call count tracking (tool_call_count) for agentic behavior metrics

  • Provider-specific usage details (usage_details dict) for reasoning tokens, etc.

  • Accurate Prompt Caching Costs

  • Cached tokens are now priced at 90% discount (Anthropic) automatically
  • CostCalculator.calculate_cost() accepts optional cache_read_tokens, cache_write_tokens parameters
  • Pricing overrides support cache token rates via add_pricing_override(cache_read_per_mtok=..., cache_write_per_mtok=...)
  • Fixes cost overreporting when prompt caching is enabled (~18% more accurate)

Changed

  • All Pydantic schemas now use Field(description=...) for better API documentation
  • ChatResponse, StepUsage, and PipelineUsage include new usage fields with sensible defaults
  • Backward compatible - existing code continues to work without changes

Documentation

  • Updated FastroAgent guide with new response fields
  • Added Prompt Caching section to Cost Calculator guide
  • Updated API reference with enhanced schemas

Full Changelog: https://github.com/benavlabs/fastroai/compare/v0.3.0...v0.4.0

[0.3.0] - Dec 17, 2025

Added

  • LogfireTracer by @igorbenav
  • Built-in tracer implementation for Pydantic's Logfire observability platform
  • Implements the Tracer protocol - drop-in replacement for SimpleTracer
  • Automatic span creation with _tags=["fastroai"] for easy filtering in Logfire dashboard
  • Metric logging via logfire.info() with trace correlation
  • Error logging with full exception info via logfire.error()

  • Clear ImportError when logfire package is not installed

  • Optional Dependency by @igorbenav

  • Install with pip install fastroai[logfire] to enable Logfire support
  • Logfire remains optional - core functionality works without it

Documentation

  • Added LogfireTracer to built-in tracers section in tracing guide
  • Added LogfireTracer to API reference
  • Updated README with Logfire installation instructions

What's Changed

Full Changelog: https://github.com/benavlabs/fastroai/compare/v0.2.0...v0.3.0

[0.2.0] - Dec 16, 2025

Added

  • FastroAgent - PydanticAI wrapper with automatic cost calculation and tracing
  • Pipeline - DAG-based workflow orchestration with automatic parallelization
  • @step decorator - Concise function-based pipeline step definitions
  • @safe_tool - Production-safe tool decorator with timeout and retry
  • CostCalculator - Precise cost tracking using microcents (integer arithmetic)
  • Tracer Protocol - Protocol-based tracing interface for observability integration
  • SimpleTracer - Logging-based tracer for development
  • NoOpTracer - No-op tracer for testing or disabled tracing

Documentation

  • Complete documentation site with guides, API reference, and recipes
  • MkDocs Material theme with dark/light mode support

Full Changelog: https://github.com/benavlabs/fastroai/compare/v0.1.0...v0.2.0

[0.1.0] - Dec 15, 2025

Initial release.

  • Core FastroAgent functionality
  • Basic pipeline support
  • Cost calculation primitives