Changelog
New updates and improvements to OpenCode
v1.3.16 Support Azure model options on both chat and responses paths
Improvements
- CoreExpose session model and mode config options through ACP@georgeharker
- CoreAdd a separating blank line before read tool file contents for clearer output
- CoreFix output token totals when reasoning tokens are reported separately
v1.3.15 v1.3.14 Restored git-backed review modes
Restored git-backed review modes, including uncommitted and branch diffs.
Improvements 13
- CoreFixed revert chains so restoring an earlier message also restores the right snapshot state.@natewill
- CoreAdded macOS managed preferences for MDM-enforced config.@lennyvaknine43
- CoreFixed sessions getting stuck after tool calls with OpenAI-compatible providers.@valenvivaldi
- CoreKept compaction summaries in the conversation's language.@aaron-he-zhu
- CoreAdded Venice AI as a provider.@dpuyosa
- CoreRespected model-specific
limit.inputoverrides.@ykswang - DesktopAdded file mentions in review comments.
- DesktopRestored prompt focus after closing agent, model, and variant pickers.
- DesktopAdded keyboard navigation and shortcuts to the question dock.
- DesktopHid model controls in shell mode.
- DesktopStopped the todo dock from auto-scrolling while tasks update.
- SDKFixed JS SDK server and TUI launch and shutdown on Windows.
- SDKFixed object-defined
Tool.define()tools from wrappingexecutemultiple times.@jpcarranza94
v1.3.12 Enabled prompt caching and cache token tracking for Google Vertex Anthropic
Improvements
- CoreFixed Azure provider options being forwarded correctly after the AI SDK v6 migration.
v1.3.11 v1.3.10 Subagent tool calls stay clickable while they are pending
Improvements
- CoreImproved storage migration reliability so malformed legacy records or failed migrations do not corrupt upgraded state.
v1.3.9 v1.3.7 Added first-class PowerShell support on Windows
Improvements
- CorePlugin installs now preserve JSONC comments in configuration files
- CoreFixed
OPENCODE_DISABLE_CLAUDE_CODE_PROMPTnot being respected for project-level CLAUDE.md
v1.3.6 PR #19758
PR #19758: Fixed token usage double-counting for Anthropic and Amazon Bedrock providers that caused inflated token counts in session metrics
v1.3.5 Fix plugin hooks to properly handle async operations
Improvements
- CoreAdjust GPT prompt to be more minimal and fix file reference annoyances