Release v1.17.0

What's Changed

• [Bug]: add NSLocalNetworkUsageDescription for local-network prompt by @a-ghorbani in #820
• Add internet search to chat (web_search + read_url talents, BYOK) by @a-ghorbani in #808
• [Feat]: Discover llama.cpp /props for remote vision + context banners by @a-ghorbani in #818
• Translations update from Hosted Weblate by @weblate in #804
• Translations update from Hosted Weblate by @weblate in #825
• [Feat]: Add Polish and European Portuguese by @a-ghorbani in #826
• chore(deps): upgrade llama.rn 0.12.4 to 0.12.7 by @a-ghorbani in #827
• [Feat]: Searchable language picker in Settings by @a-ghorbani in #829
• fix: make markdown link color theme-aware (#753) by @sakebomb in #834
• Translations update from Hosted Weblate by @weblate in #832
• [Bug]: scope the llama.cpp /props capability probe to a model by @a-ghorbani in #830
• Translations update from Hosted Weblate by @weblate in #846
• feat: export a chat session as Markdown by @sakebomb in #836
• feat: add pinned/starred chat sessions by @sakebomb in #621
• Translations update from Hosted Weblate by @weblate in #851
• [Feat]: Speculative decoding / MTP draft-model support by @a-ghorbani in #805
• fix: stop local models disappearing when the app container path changes by @sakebomb in #835
• chore(deps): upgrade llama.rn to 0.13.0-rc.0 by @a-ghorbani in #857
• Translations update from Hosted Weblate by @weblate in #852

Full Changelog: v1.16.1...v1.17.0

Release v1.16.1

What's Changed

• [Feat]: Android in-app PalsHub checkout + Play External Content Links by @a-ghorbani in #766
• [Bug]: Fix PalsHub profile load permission error (narrow profiles select) by @a-ghorbani in #802
• [Bug]: gate PalsHub buy button on real purchase eligibility, not device locale by @a-ghorbani in #814

Full Changelog: v1.16.0...v1.16.1

Release v1.16.0

What's Changed

• [Bug]: E2E fixes from v1.15.2 release validation (#764) by @a-ghorbani in #765
• Translations update from Hosted Weblate by @weblate in #726
• [Bug]: Stopping a download no longer shows "Download Failed" dialog by @a-ghorbani in #771
• [Feat]: Configurable per-server network request timeout for remote models by @a-ghorbani in #778
• [Bug]: Keep latest AI output visible when Android keyboard opens by @a-ghorbani in #780
• docs: update README by @a-ghorbani in #781
• docs: add on-device stack diagram by @a-ghorbani in #782
• [Feat]: Device-rule-driven default model list by @a-ghorbani in #772
• feat(onboarding): add splash, 6-step stack, and Pip pal seed by @a-ghorbani in #747
• feat(tts): Supertonic 31-language synthesis + language picker by @a-ghorbani in #784
• [Feat]: Reasoning model detection + user override (local + remote) by @a-ghorbani in #783
• Translations update from Hosted Weblate by @weblate in #769
• Translations update from Hosted Weblate by @weblate in #800
• Repair stale/flaky e2e feature specs (post #772/#783) by @a-ghorbani in #798
• [Feat]: Add Brazilian Portuguese (pt-BR) language support by @a-ghorbani in #801

Full Changelog: v1.15.2...v1.16.0

Release v1.15.2

What's Changed

• [Bug]: PalsHub email sign-up/sign-in no longer claims success on failure by @a-ghorbani in #757
• [Feat]: Open Hugging Face model downloads via hub/run deep link by @a-ghorbani in #761
• [Feat]: Warn when a chat is near or at the context limit, with recovery by @a-ghorbani in #763
• [Feat]: Authenticated in-app PalsHub checkout (US iOS) by @a-ghorbani in #752

Full Changelog: v1.15.1...v1.15.2

Release v1.15.1

What's Changed

• [Feat]: non-touch autostart deep link for bench E2E matrix by @a-ghorbani in #733
• [Bug]: pin fmt pod to C++17 to fix clang-21 (Xcode 26.4.1) iOS CI builds by @a-ghorbani in #739
• [Feat]: design-token foundation + fonts + theme decoupling by @a-ghorbani in #732
• chore(deps): upgrade llama.rn to 0.12.3 by @a-ghorbani in #740
• chore(deps): upgrade llama.rn to 0.12.4 by @a-ghorbani in #743
• [Bug]: pal card falls back to creator avatar when no thumbnail by @a-ghorbani in #746
• [Feat]: wire Palshub pact.talents + greeting into pal download by @a-ghorbani in #741
• [Bug]: thinking toggle locked by pal in no-session chat path by @a-ghorbani in #745
• [Feat]: shared UI component library (src/components/ui/) by @a-ghorbani in #742

Full Changelog: v1.15.0...v1.15.1

Release v1.15.0

What's Changed

• Translations update from Hosted Weblate by @weblate in #683
• [Bug]: Move iOS workflows to macOS 26 by @a-ghorbani in #727
• chore(ci): rework e2e-tests into ref-targeted Android build-only workflow by @a-ghorbani in #729
• chore(ci): drop redundant workflow_dispatch ref input from e2e-tests by @a-ghorbani in #731
• chore(deps): upgrade llama.rn to 0.12.1 by @a-ghorbani in #728
• feat(chat): introduce structured assistant turns for tool calling by @a-ghorbani in #709
• fix(chat): tighten WebView JS surface + interrupted turn UX by @a-ghorbani in #737

Full Changelog: v1.14.1...v1.15.0

Release v1.14.1

What's Changed

• fix(release): increase Node heap size for iOS bundle step by @a-ghorbani in #696
• feat(release): add ios_only dispatch input by @a-ghorbani in #697
• chore(ci): retain iOS dSYM as release workflow artifact by @a-ghorbani in #701
• feat(e2e): benchmark-matrix infra + automation-bridge architecture + BenchmarkRunnerScreen by @a-ghorbani in #702
• chore(deps): bump react-native-speech 2.2.2 → 2.3.1 by @a-ghorbani in #719
• feat(bench): settings sweep + Hexagon backend (v1.1 schema) by @a-ghorbani in #713
• feat(settings): add TTS availability toggle + fix Kokoro silent-audio overflow by @a-ghorbani in #717
• feat(android): enable Adreno large buffer for A7X/A8X GPUs by @a-ghorbani in #699
• chore(deps): upgrade llama.rn to 0.12.0 by @a-ghorbani in #722

Full Changelog: v1.14.0...v1.14.1

Release v1.14.0

What's Changed

• Upgrade react-native-keyboard-controller to fix iOS startup crash by @a-ghorbani in #678
• feat(models): improve local model upload flow visibility by @a-ghorbani in #676
• Translations update from Hosted Weblate by @weblate in #647
• feat(l10n): add Ukrainian and Chinese Traditional languages by @a-ghorbani in #672
• Fix completion settings inconsistencies with pal overrides by @a-ghorbani in #681
• feat(pal): add combobox input type for dynamic parameter forms by @a-ghorbani in #682
• Rewrite Play Store listing by @a-ghorbani in #685
• feat(settings): lift n_predict limit, default to unlimited by @a-ghorbani in #687
• chore(deps): upgrade llama.rn to 0.12.0-rc.8 by @a-ghorbani in #689
• feat(tts): neural TTS with streaming, 4 engines, voice-led setup by @a-ghorbani in #688

Full Changelog: v1.13.3...v1.14.0

Release v1.13.3

What's Changed

• Fix app crash loop when loadLlamaModelInfo fails by @a-ghorbani in #671

Full Changelog: v1.13.2...v1.13.3

Release v1.13.2

What's Changed

• chore(deps): upgrade llama.rn to 0.12.0-rc.3 by @a-ghorbani in #664
• Replace manual thinking detection with llama.rn getFormattedChat by @a-ghorbani in #666

Full Changelog: v1.13.1...v1.13.2