Skip to content

monolabsdev/openbench-ai

v0.20.0 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 14d LLM Frameworks
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

local-llm multi-model offline-ai ollama privacy-first tauri

Summary

AI summary

Updates affect Rust build, Linux window resize, voice mode/settings, Memory beta feature, and performance tweaks for TTS and shaders.

Changes in this release

Feature Medium

Adds Memory (Beta) settings tab and toggle.

Adds Memory (Beta) settings tab and toggle.

Source: llm_adapter@2026-07-13

Confidence: high

Feature Medium

Adds voice mode updates and related settings.

Adds voice mode updates and related settings.

Source: llm_adapter@2026-07-13

Confidence: high

Feature Medium

Shows offline provider notice in chat empty state.

Shows offline provider notice in chat empty state.

Source: llm_adapter@2026-07-13

Confidence: high

Feature Low

Rate‑limits failed token authentication during pairing.

Rate‑limits failed token authentication during pairing.

Source: granite4.1:30b@2026-07-13-audit

Confidence: high

Performance Medium

Releases TTS engine on idle and pauses orb shader.

Releases TTS engine on idle and pauses orb shader.

Source: llm_adapter@2026-07-13

Confidence: high

Performance Medium

Lazy‑loads agent viewport drawer for better performance.

Lazy‑loads agent viewport drawer for better performance.

Source: llm_adapter@2026-07-13

Confidence: high

Bugfix Medium

Fixes window edge resizing on Linux.

Fixes window edge resizing on Linux.

Source: llm_adapter@2026-07-13

Confidence: high

Bugfix Medium

Fixes settings deep‑merge defaults on rehydrate.

Fixes settings deep‑merge defaults on rehydrate.

Source: llm_adapter@2026-07-13

Confidence: high

Bugfix Medium

Fixes surface swallowed errors to show as toast notifications.

Fixes surface swallowed errors to show as toast notifications.

Source: llm_adapter@2026-07-13

Confidence: high

Bugfix Medium

Fixes ONNX Runtime linux arm64 asset name in build.

Fixes ONNX Runtime linux arm64 asset name in build.

Source: llm_adapter@2026-07-13

Confidence: high

Bugfix Medium

Fixes reliable memory extraction and persistent chip handling.

Fixes reliable memory extraction and persistent chip handling.

Source: llm_adapter@2026-07-13

Confidence: low

Bugfix Low

Surfaces previously swallowed errors as toast notifications.

Surfaces previously swallowed errors as toast notifications.

Source: granite4.1:30b@2026-07-13-audit

Confidence: high

Refactor Low

Renames voice profiles to neutral names.

Renames voice profiles to neutral names.

Source: granite4.1:30b@2026-07-13-audit

Confidence: high

Full changelog
  • Merge pull request #124 from monolabsdev/fix/onnx-arm64-url (01634cb)
  • fix(build): correct ONNX Runtime linux arm64 asset name (34dcce9)
  • fix(rust): unwrap read_dir Result entries in build.rs (07bec04)
  • fix(rust): download ONNX Runtime in build.rs when using load-dynamic (8f9eb1d)
  • fix(rust): import CommandExt for creation_flags on Windows (fcc3230)
  • Merge pull request #123 from monolabsdev/chore/clippy-cleanup (3493e0c)
  • chore(rust): clear all clippy warnings (274cb43)
  • Merge pull request #122 from monolabsdev/fix/linux-window-resize (7fee082)
  • fix(linux): restore window edge resizing (dca6931)
  • Merge pull request #121 from monolabsdev/chore/v0.20.0 (c686026)
  • refactor(voice): rename voice profiles to neutral names (6296d21)
  • chore: bump version to v0.20.0 (f60a02c)
  • Merge pull request #119 from monolabsdev/feat/memory-beta (745e23d)
  • Merge pull request #120 from monolabsdev/perf/idle-tts-orb (e2464bd)
  • perf: release TTS engine on idle, pause orb shader (7c02abc)
  • fix(memory): reliable extraction and persistent chip (c3d54df)
  • feat(settings): Memory (Beta) toggle and settings tab (4c8d107)
  • Merge pull request #117 from monolabsdev/feat/v0.20-polish (5887006)
  • chore(chat): drop stale prompt-kit migration TODO (b0fd56e)
  • perf: lazy-load agent viewport drawer (438b94f)
  • feat(pairing): rate-limit failed token auth (acccb7f)
  • refactor(voice): use theme tokens for overlay colors (7a05e65)
  • fix(settings): deep-merge defaults on every rehydrate (d7708fe)
  • fix: surface swallowed errors as toasts (9706cdf)
  • feat(chat): show offline provider notice in empty state (6d9854d)
  • Merge pull request #116 from monolabsdev/feat/voice-mode-and-settings-updates (f9d7b56)
  • fix: use explicit hex colors in voice overlay to satisfy tests (c7399f0)
  • feat: voice mode and settings updates (4e5521f)
  • docs: add auto-update relaunch design spec (f3a6da9)

What's Changed

  • feat: voice mode and settings updates by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/116
  • v0.20.0 polish: offline state, error toasts, pairing rate limit, settings healing by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/117
  • perf: idle-aware TTS engine release and orb shader pause by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/120
  • Memory (Beta): settings tab, reliable extraction, persistent chip by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/119
  • chore: prepare v0.20.0 release by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/121
  • fix(linux): restore window edge resizing by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/122
  • chore(rust): clear all clippy warnings by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/123
  • fix(build): correct ONNX Runtime linux arm64 asset name by @TheoSlater in https://github.com/monolabsdev/poly-ui/pull/124

Full Changelog: https://github.com/monolabsdev/poly-ui/compare/v0.19.3...v0.20.0

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track monolabsdev/openbench-ai

Get notified when new releases ship.

Sign up free

About monolabsdev/openbench-ai

All releases →

Related context

Beta — feedback welcome: [email protected]