Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog. This project uses Calendar Versioning with the format YYYY.MM.PATCH.

2026.2.1 (2026-02-01)

Features:

• Add pricing and billing support for tool usage (#92) by Victor

• Initial billing support for tool usage by Victor

• add metadata and pricing for OpenAI image generation models by Victor

• add Pricing and Billing guide by Victor

• Support model specific base_url (#86) by meanderingstream

• add model specific base_url configuration by meanderingstream

• Add documentation for model base_url by meanderingstream

• Add Cohere models to Bedrock (#90) by ycastorium

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 3 provider(s) (#104) by github-actions[bot]

• proper metadata for grok-imagine-image (#103) by Victor

• add xai image model metadata by Victor

• correct grok-imagine-image pricing by Victor

• add grok-imagine-image input image pricing by Victor

• update model metadata for 17 provider(s) (#102) by github-actions[bot]

• add explicit wire.protocol metadata for OpenAI models (#101) by Victor

• explicitly define metadata for modern OpenAI models, fix #100 by Victor

• test environment (#98) by Victor

• clear config before each test by Victor

• allow provider names to start with a digit (302ai is a legitimate provider) by Victor

• update model metadata for 3 provider(s) (#97) by github-actions[bot]

• add pricing metadata, close #93 (#96) by Victor

• add pricing information for gemini-3-pro-image-preview by Victor

• add pricing for gpt-3.5-turbo-16k by Victor

• update model metadata for 84 provider(s) (#94) by github-actions[bot]

• remove cost.image to pricing component conversion by Victor

• correct Google cache and embedding pricing by Victor

• handle string id_key for atom maps by Victor

• clear :filter env in test setup to prevent pollution by Victor

• Elixir 1.20.0-rc.1 warnings by Victor

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.2.0 (2026-02-01)

Features:

• Add pricing and billing support for tool usage (#92) by Victor

• Initial billing support for tool usage by Victor

• add metadata and pricing for OpenAI image generation models by Victor

• add Pricing and Billing guide by Victor

• Support model specific base_url (#86) by meanderingstream

• add model specific base_url configuration by meanderingstream

• Add documentation for model base_url by meanderingstream

• Add Cohere models to Bedrock (#90) by ycastorium

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• proper metadata for grok-imagine-image (#103) by Victor

• add xai image model metadata by Victor

• correct grok-imagine-image pricing by Victor

• add grok-imagine-image input image pricing by Victor

• update model metadata for 17 provider(s) (#102) by github-actions[bot]

• add explicit wire.protocol metadata for OpenAI models (#101) by Victor

• explicitly define metadata for modern OpenAI models, fix #100 by Victor

• test environment (#98) by Victor

• clear config before each test by Victor

• allow provider names to start with a digit (302ai is a legitimate provider) by Victor

• update model metadata for 3 provider(s) (#97) by github-actions[bot]

• add pricing metadata, close #93 (#96) by Victor

• add pricing information for gemini-3-pro-image-preview by Victor

• add pricing for gpt-3.5-turbo-16k by Victor

• update model metadata for 84 provider(s) (#94) by github-actions[bot]

• remove cost.image to pricing component conversion by Victor

• correct Google cache and embedding pricing by Victor

• handle string id_key for atom maps by Victor

• clear :filter env in test setup to prevent pollution by Victor

• Elixir 1.20.0-rc.1 warnings by Victor

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.5 (2026-01-30)

Features:

• Add pricing and billing support for tool usage (#92) by Victor

• Initial billing support for tool usage by Victor

• add metadata and pricing for OpenAI image generation models by Victor

• add Pricing and Billing guide by Victor

• Support model specific base_url (#86) by meanderingstream

• add model specific base_url configuration by meanderingstream

• Add documentation for model base_url by meanderingstream

• Add Cohere models to Bedrock (#90) by ycastorium

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 17 provider(s) (#102) by github-actions[bot]

• add explicit wire.protocol metadata for OpenAI models (#101) by Victor

• explicitly define metadata for modern OpenAI models, fix #100 by Victor

• test environment (#98) by Victor

• clear config before each test by Victor

• allow provider names to start with a digit (302ai is a legitimate provider) by Victor

• update model metadata for 3 provider(s) (#97) by github-actions[bot]

• add pricing metadata, close #93 (#96) by Victor

• add pricing information for gemini-3-pro-image-preview by Victor

• add pricing for gpt-3.5-turbo-16k by Victor

• update model metadata for 84 provider(s) (#94) by github-actions[bot]

• remove cost.image to pricing component conversion by Victor

• correct Google cache and embedding pricing by Victor

• handle string id_key for atom maps by Victor

• clear :filter env in test setup to prevent pollution by Victor

• Elixir 1.20.0-rc.1 warnings by Victor

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.4 (2026-01-28)

Features:

• Add pricing and billing support for tool usage (#92) by Victor

• Initial billing support for tool usage by Victor

• add metadata and pricing for OpenAI image generation models by Victor

• add Pricing and Billing guide by Victor

• Support model specific base_url (#86) by meanderingstream

• add model specific base_url configuration by meanderingstream

• Add documentation for model base_url by meanderingstream

• Add Cohere models to Bedrock (#90) by ycastorium

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• test environment (#98) by Victor

• clear config before each test by Victor

• allow provider names to start with a digit (302ai is a legitimate provider) by Victor

• update model metadata for 3 provider(s) (#97) by github-actions[bot]

• add pricing metadata, close #93 (#96) by Victor

• add pricing information for gemini-3-pro-image-preview by Victor

• add pricing for gpt-3.5-turbo-16k by Victor

• update model metadata for 84 provider(s) (#94) by github-actions[bot]

• remove cost.image to pricing component conversion by Victor

• correct Google cache and embedding pricing by Victor

• handle string id_key for atom maps by Victor

• clear :filter env in test setup to prevent pollution by Victor

• Elixir 1.20.0-rc.1 warnings by Victor

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.3 (2026-01-28)

Features:

• Add pricing and billing support for tool usage (#92) by Victor

• Initial billing support for tool usage by Victor

• add metadata and pricing for OpenAI image generation models by Victor

• add Pricing and Billing guide by Victor

• Support model specific base_url (#86) by meanderingstream

• add model specific base_url configuration by meanderingstream

• Add documentation for model base_url by meanderingstream

• Add Cohere models to Bedrock (#90) by ycastorium

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 84 provider(s) (#94) by github-actions[bot]

• remove cost.image to pricing component conversion by Victor

• correct Google cache and embedding pricing by Victor

• handle string id_key for atom maps by Victor

• clear :filter env in test setup to prevent pollution by Victor

• Elixir 1.20.0-rc.1 warnings by Victor

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.2 (2026-01-26)

Features:

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 20 provider(s) (#85) by github-actions[bot]

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.1 (2026-01-19)

Features:

• metadata: add Zenmux API key to build metadata workflow by mikehostetler

• providers: add support for Zenmux provider (#75) by youfun

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 67 provider(s) (#84) by github-actions[bot]

• add metadata for grok-4-0709 and grok-4-1-fast-reasoning (#83) by Victor

• Add mising metadata for grok-4-0709 and grok-4-1-fast-reasoning by Victor

• update model metadata for 8 provider(s) (#79) by github-actions[bot]

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2026.1.0 (2026-01-05)

Features:

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 13 provider(s) (#72) by github-actions[bot]

• keep OpenRouter models under :openrouter with full IDs (#70) by mikehostetler

• update model metadata for 17 provider(s) (#71) by github-actions[bot]

• update model metadata for 15 provider(s) (#65) by github-actions[bot]

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2025.12.4 (2025-12-26)

Features:

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• Don't load env files if .env is a directory (#64) by sezaru

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2025.12.3 (2025-12-22)

Features:

• add wire.protocol and constraints metadata fields (#59) by mikehostetler

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 12 provider(s) (#61) by github-actions[bot]

• inherit custom config from app env in Runtime.compile/1 (#58) by Nils

• update model metadata for 51 provider(s) (#57) by github-actions[bot]

• round cost values to six decimal places in OpenRouter by mikehostetler

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

Improvements:

• Add gemini-2.5-flash-image and gemini-2.5-flash-lite (#60) by Victor

2025.12.2 (2025-12-17)

Features:

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• update model metadata for 5 provider(s) (#55) by github-actions[bot]

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

2025.12.1 (2025-12-17)

Features:

• add hex_dry_run option and version bump commit to release workflow by mikehostetler

• sort JSON keys for deterministic output (#51) by mikehostetler

• sort JSON keys for deterministic output by mikehostetler

• add git_hooks integration for code quality enforcement (#42) by mikehostetler

Bug Fixes:

• add fetch-tags option and debug output to release workflow by mikehostetler

• fetch tags explicitly in release workflow by mikehostetler

• update model metadata for 69 provider(s) (#50) by github-actions[bot]

• disable git hooks in CI workflow by mikehostetler

• use commit-message instead of invalid commit-message-path by mikehostetler

• restrict llm_db.build and llm_db.pull tasks to llm_db project only (#49) by mikehostetler

2025.12.0 - 2025-12-17

Changed

• Updated model data from upstream sources (includes Gemini 3 Pro Preview)
• JSON snapshot keys are now sorted for deterministic output (#51)
• Added git_hooks integration for code quality enforcement (#42)
• Restricted llm_db.build and llm_db.pull tasks to llm_db project only (#49)

[2025.11.18-preview] - 2025-11-18

Added

• New models available: GPT 5.1 (OpenAI) and Gemini 3 (Google)
• Provider alias system to enable single implementation handling models from multiple LLMDB providers
  • New alias_of field in Provider struct points aliased provider to primary implementation
  • LLMDB.Store.models/1 now searches aliased providers when looking up by provider ID
  • LLMDB.Store.model/2 normalizes provider field back to requested provider for aliased models
  • First implementation: google_vertex_anthropic aliases to google_vertex for Claude models on Vertex AI
• provider_model_id field for AWS Bedrock inference profile models that require API-specific identifiers
  • Enables models to use canonical IDs (e.g., anthropic.claude-haiku-4-5-20251001-v1:0) while making API calls with inference profile prefixes (e.g., us.anthropic.claude-haiku-4-5-20251001-v1:0)
  • Addresses AWS requirement: "Invocation of model ID [...] with on-demand throughput isn't supported. Retry your request with the ID or ARN of an inference profile"
  • Applied to: Claude Haiku 4.5, Claude Sonnet 4.5, Claude Opus 4.1, Llama 3.3 70B, Llama 3.2 3B
• .env.example file for environment variable configuration

Changed

• Snapshot JSON is now pretty-printed for easier diffing
• Updated Zoi dependency to version 0.10.7
• Updated Dotenvy dependency to version 1.1
• ModelsDev transformer now auto-sets streaming.tool_calls: true when tool_call: true
  • Reflects reality: 99%+ of tool-capable models support streaming tool calls
  • Eliminates need for model-specific TOML overrides for common case
  • Rare exceptions can override with streaming.tool_calls: false in TOML

Fixed

• Claude Opus 4.1: Changed provider_model_id from global. to us. prefix - Opus 4.1 is only available with us. inference profile on AWS Bedrock, not global. like Haiku and Sonnet
• Claude Haiku 4.5 and Sonnet 4.5: Override tools.strict=false to disable object generation hack - waiting for native Anthropic JSON support instead
• Model spec parsing now handles ambiguous formats (specs with both : and @ separators) by attempting provider validation to determine the correct format
• Removed overly strict character validation that rejected @ in model IDs when using colon format and : in model IDs when using @ format

[2025.11.14-preview] - 2025-11-14

Added

• LLMDB.Model.format_spec/1 function for converting model struct to provider:model string format
• Zai Coder provider and GLM models support
• Enhanced cost schema with granular multimodal and reasoning pricing fields:
  • reasoning: Cost per 1M reasoning/thinking tokens for models like o1, Grok-4
  • input_audio/output_audio: Separate audio input/output costs (e.g., Gemini 2.5 Flash, Qwen-Omni)
  • input_video/output_video: Video input/output cost support for future models
• ModelsDev source transformer now captures all cost fields from models.dev data
• OpenRouter source transformer maps internal_reasoning field to reasoning cost

Changed

• Updated Zoi dependency to version 0.10.6
• Refactored loader to use dynamic snapshot retrieval
• Disabled schema validation in snapshot.json and TOML source files

Fixed

• Cleaned up code quality issues
• Fixed application startup crash (ArgumentError: not an already existing atom) caused by race condition between build task and snapshot loading
• Fixed flaky tests in LLMDB.EngineOverrideTest by ensuring test isolation from global config