v0.1.6a1

[0.1.6a1] - 2026-03-24

Added

• OpenAI Files API: added /files, /files/{file_id}, and /files/{file_id}/content endpoints plus the examples/openai/files.py example
• OpenAI Batches API: added /batches and /batches/{batch_id} endpoints plus the examples/openai/batches.py example
• Anthropic Message Batches API: added /v1/messages/batches, /v1/messages/batches/{message_batch_id}, and /v1/messages/batches/{message_batch_id}/results plus the examples/anthropic/message_batches.py example
• New integrations: added setup guides for Qwen Code and Xcode
• CI and automation: added actionlint, CodeQL, dependency-review, docker-smoke, nightly-smoke, pr-labeler, release-drafter, stale-issues, and Dependabot configuration

Changed

• Examples: moved OpenAI examples under examples/openai/ and aligned README/AGENTS docs with the new layout
• OpenAPI: split OpenAI and Anthropic schema builders into gpt2giga/openapi_specs/
• LiteLLM router: moved /model/info handling into the dedicated gpt2giga/routers/litellm/ package
• Docker Compose: standardized compose files under compose/ (base.yaml, observability.yaml, nginx.yaml, observe-multiple.yaml, traefik.yaml)
• GitHub templates: added Russian-language issue and pull request templates

Fixed

• Path normalization: fixed normalization for /v1, files, batches, messages, and model/info
• OpenAI payload mapping: extra_body now maps correctly to additional_fields
• Batches: fixed completion_window handling and Python 3.10 datetime behavior
• Examples: refreshed runnable OpenAI and Anthropic examples after the directory reorganization

v0.1.5

What's Changed

• feature: add /model/info for compatability with kilocode autocompleti… by @krakenalt in #80

Added

• Model info endpoint: Added GET /model/info for compatibility with Kilo Code autocomplete and LiteLLM-style clients

Changed

• GitHub Actions: ci.yaml, docker_image.yaml, and publish-ghcr.yml now run only when relevant files change

Fixed

• Pull Request CI: The test workflow no longer runs for draft PRs

Full Changelog: v0.1.4.post1...v0.1.5

v0.1.4.post1

What's Changed

• Development environment setup by @krakenalt in #78

Full Changelog: v0.1.4...v0.1.4.post1

Added

• Cursor integration: Added integrations/cursor/README.md — guide for using GigaChat in Cursor as a custom model
• Codex integration: Added integrations/codex/README.md — OpenAI Codex setup via config.toml with custom gpt2giga provider
• Claude Code integration: Added integrations/claude-code/README.md — Claude Code setup via ANTHROPIC_BASE_URL
• AGENTS.md documentation: Updated all AGENTS.md files to match the current codebase structure

Changed

• Async I/O: Moved blocking I/O operations in route handlers to worker threads via anyio.to_thread.run_sync:
  • logs_router.py — log file reading and HTML template loading
  • api_router.py — tiktoken.encoding_for_model() initialization

v0.1.4

What's Changed

• Relese v0.1.4 by @krakenalt in #76

Full Changelog: v0.1.3.post1...v0.1.4

Added

• Nginx: Added gpt2giga.conf config and README for deployment behind nginx
• Docker Compose: Updated compose (#77) — mitmproxy in docker-compose-observability.yaml, password for mitmproxy
• Logs router: Extracted logs_router.py, split system router in two

Changed

• Updated .env.example
• Updated README for nginx

Fixed

• Giga-auth: Fixed giga-auth behaviour (#74)

v0.1.3.post1

Added

• Traefik: Added Traefik integration (#70) by @2100992
• MITMProxy: Added mitmproxy to docker-compose-observability.yaml
• Reasoning toggle: Added GPT2GIGA_ENABLE_REASONING environment variable

Changed

• Docker Compose profiles: Set dev as the default profile

New Contributors

• @2100992 made their first contribution in #70

v0.1.3

Added

• DEV/PROD Mode: Added support for development and production modes
• Configurable CORS: Added CORS configuration via environment variables
• Graceful shutdown: Added graceful server shutdown handling
• Gitleaks: Added gitleaks to pre-commit for secret detection
• OpenAPI for count_tokens: Added OpenAPI documentation for count_tokens endpoint
• • Profiles in Docker: Added profiles DEV and PROD in docker-compose.yaml

Changed

• Structure Refactoring: Split large files into modules:
  • gpt2giga/common/ — common utilities (exceptions, json_schema, streaming, tools)
  • gpt2giga/models/ — configuration and security models
  • gpt2giga/protocol/attachment/ — attachment processing
  • gpt2giga/protocol/request/ — request transformation
  • gpt2giga/protocol/response/ — response processing
• Improved Logging: Log redaction policy, disabled full payload logging

Fixed

• CLI Security: Fixed command-line argument issues
• Port Bindings: Fixed port binding and redirect issues
• SSRF Protection: Hardened SSRF protection in attachment handling
• Authentication: Switched to secrets.compare_digest for key comparison
• Attachment Limits: Added limits for attachments
• Mapping for reversed tool name: Fixed bug with function name web_search, which can break function_call

v0.1.2.post1

Added

• OpenAPI Documentation: Added full OpenAPI documentation for all endpoints
• Count tokens for Anthropic: Added /v1/messages/count_tokens endpoint for token counting in Anthropic format
• count_tokens Example: Added examples/anthropic/count_tokens.py example
• Version on Initialize: Display version on server startup

Changed

• Path Normalizer: Improved path normalizer for responses and messages

Fixed

• 405 Error: Fixed 405 error on certain requests
• Safe Request Reading: Improved request body reading handling

v0.1.2

What's Changed

• Release v0.1.2 by @krakenalt in #64

Full Changelog: v0.1.1.post1...v0.1.2

Added

• Anthropic Messages API: New POST /v1/messages endpoint for Anthropic Messages API compatibility
  • Streaming support (SSE) in Anthropic format (message_start, content_block_delta, message_stop, etc.)
  • Anthropic message conversion (text, image, tool_use, tool_result) to GigaChat format
  • Anthropic tool conversion (input_schema) to GigaChat functions format
  • tool_choice support (auto, tool, none)
  • System prompt support (system) as string or content block array
  • stop_reason mapping (end_turn, tool_use, max_tokens)
• Extended Thinking (Reasoning): Support for Anthropic thinking parameter
  • thinking.budget_tokens mapped to GigaChat reasoning_effort (high/medium/low)
  • GigaChat reasoning_content converted to Anthropic thinking content block
  • Reasoning support in streaming (thinking_delta)
• Anthropic API Examples: Added examples in examples/anthropic/:
  • messages.py — basic request
  • messages_stream.py — streaming
  • system_prompt.py — system prompt
  • multi_turn.py — multi-turn conversation
  • function_calling.py — function calling (tool use)
  • image_url.py — image from URL
  • base64_image.py — base64 image
  • reasoning.py — extended thinking

v0.1.1.post1

What's Changed

• bugfix: Handle array-style type field: type: ['string', 'null'] -> ty… by @krakenalt in #61

Full Changelog: v0.1.1...v0.1.1.post1

v0.1.1

What's Changed

• [Release]: v0.1.1 by @krakenalt in #59

Full Changelog: v0.1.0...v0.1.1

Added

• GitHub Templates: Added Pull Request and Issue (bug report) templates to improve contribution process (#58)
• $ref Resolution in Schemas: Added _resolve_schema_refs function for handling JSON Schema references (#57)
• Missing Properties Handling: Implemented proper handling of schemas without properties field

Changed

• request_mapper.py Refactoring: Logic split into separate modules for better maintainability:
  • content_utils.py — utilities for content handling
  • message_utils.py — utilities for message handling
  • schema_utils.py — utilities for schema handling
• Extended Test Coverage: Added tests for streaming and tool conversion

Fixed

• Responses API Streaming: Fixed streaming responses in Responses API (#60)
• Function Calling in Streaming: Fixed function call handling during streaming in Responses API