2.0.64

• Localize log pagination and tighten packaging tests
• fix: 修复缓存统计和 WebSearch 转换 (#392)
• fix: sync log controls and preserve zero-latency display
• feat: separate logs settings drawer and sync enabled state
• feat: add persisted LLM-only log capture mode
• fix: correct logs pagination and honor body capture setting
• fix: tighten logs pagination state handling
• feat: add dashboard logs tab
• feat: 本地化 logs 的相关设置项
• refactor: bound request logs and add coverage
• feat: add dashboard logs tab
• fix
• feat: add log tab
• feat: add logs tab and backend
• test: 添加简易测试 test:local-chat 只覆盖了各个模型的 openai 格式的 curl 请求
• test: 添加简易测试 test:local-chat 只覆盖了各个模型的 openai 格式的 curl 请求

2.0.62

• refactor: open tests/ and consolidate test directory structure (#373)

2.0.61

• fix: codex models return 404 when upstream providers are configured (#369)

2.0.60

• fix: remove redundant proxy assignment header button (#363)

2.0.59

Bug fixes and improvements

2.0.57

Bug fixes and improvements

2.0.56

• fix: expose runtime API key models in model list
• fix: new account with null window_reset_at no longer starved by least_used strategy (#342) (#346)

2.0.55

• feat: add trust_proxy to fix dashboard auth bypass via tunnel (#351)

2.0.54

• fix: make reasoning effort opt-in, disable by default (#345)

2.0.53

• fix: handle per-model errors in addEntries without failing entire batch
• feat: fetch models in the custom provider
• Create codex-tls.win32-x64-msvc.node
• feat: support multiple models per API key
• feat: support multiple models per API key
• feat: support multiple models per API key