jiroamato · Oct 27, 2026 · Oct 22, 2026 · Oct 22, 2026 · Oct 22, 2026 · Oct 22, 2026
diff --git a/README.md b/README.md
@@ -179,10 +179,6 @@ PackyCode provides special discounts for our software users: register using <a h
 <td>Bestproxy provides high-purity residential IPs with dedicated one-IP-per-account support. 🟡Residential Proxy - &#36;0.5/GB；🟡Static Residential Proxy - Starting at &#36;3/IP；🟡Unlimited Residential Proxy - Starting at &#36;67/Day.  ✅<a href="https://bestproxy.com/?keyword=ayh7otlb">Get Free Trial.</a></td>
 </tr>
 <tr>
-<td width="180"><a href="https://go.apimart.ai/gh-cliproxyapi"><img src="./assets/apimart-en.png" alt="APIMart" width="150"></a></td>
-<td>Thanks to APIMart for sponsoring this project! APIMart is a low-cost API platform for AI image &amp; video generation — GPT-Image-2 from &#36;0.006/image, 160+ images per dollar. One async API covers both image and video: submit a task, get an ID, fetch results via polling or callback. Batch tens of thousands of images without timeouts, switch models without changing code. Pay-as-you-go with no monthly fee — <a href="https://go.apimart.ai/gh-cliproxyapi">sign up here</a> to get started.</td>
-</tr>
-<tr>
 <td width="180"><a href="https://www.swiftproxy.net/?code=PR67S9A95"><img src="./assets/swiftproxy.png" alt="Swiftproxy" width="150"></a></td>
 <td>Swiftproxy provides 90M+ clean residential IPs across 220+ locations, supporting HTTP(S)/SOCKS5, IP rotation, Sticky Sessions, and precise location targeting. It helps AI API tools and automation workflows access online services reliably from different locations, making it ideal for API requests, web access, data collection, and location-based testing. Residential proxies from &#36;0.7/GB. Free testing is available, with 10% off using code PROXY90. <a href="https://www.swiftproxy.net/?code=PR67S9A95">Try Swiftproxy Now</a></td>
 </tr>
@@ -305,14 +301,6 @@ Native macOS menu bar app that unifies Claude, Gemini, OpenAI, and Antigravity s
 
 Windows-native CLIProxyAPI fork with TUI, system tray, and multi-provider OAuth for AI coding tools - no API keys needed.
 
-### [Claude Proxy VSCode](https://github.com/uzhao/claude-proxy-vscode)
-
-VSCode extension for quick switching between Claude Code models, featuring integrated CLIProxyAPI as its backend with automatic background lifecycle management.
-
-### [ZeroLimit](https://github.com/0xtbug/zero-limit)
-
-Windows desktop app built with Tauri + React for monitoring AI coding assistant quotas via CLIProxyAPI. Track usage across Gemini, Claude, OpenAI Codex, and Antigravity accounts with real-time dashboard, system tray integration, and one-click proxy control - no API keys needed.
-
 ### [CPA-XXX Panel](https://github.com/ferretgeek/CPA-X)
 
 A lightweight web admin panel for CLIProxyAPI with health checks, resource monitoring, real-time logs, auto-update, request statistics and pricing display. Supports one-click installation and systemd service.

diff --git a/README_CN.md b/README_CN.md
@@ -80,10 +80,6 @@ PackyCode 为本软件用户提供了特别优惠：使用<a href="https://www.p
 <td>感谢 Cubence 对本项目的赞助！Cubence 是一家可靠高效的 API 中转服务商，提供 Claude Code、Codex、Gemini 等多种服务的中转。Cubence 为本软件用户提供了特别优惠：使用<a href="https://cubence.com/signup?code=CLIPROXYAPI&source=cpa">此链接</a>注册，并在充值时输入 "CLIPROXYAPI" 优惠码即可享受九折优惠。</td>
 </tr>
 <tr>
-<td width="180"><a href="https://go.apimart.ai/gh-cliproxyapi"><img src="./assets/apimart-zh.png" alt="APIMart" width="150"></a></td>
-<td>感谢 APIMart 赞助了本项目！APIMart 是专注 AI 图片/视频生成的低价 API 平台，GPT-Image-2 低至 &#36;0.006/张，1 美元可出图 160+ 张。图片、视频一套异步 API 通吃，提交任务拿 ID、回调取结果，跑批万张不超时、换模型不改代码。按量付费、无月费，通过<a href="https://go.apimart.ai/gh-cliproxyapi">此注册链接</a>注册即可开用。</td>
-</tr>
-<tr>
 <td width="180"><a href="https://www.swiftproxy.net/?code=PR67S9A95"><img src="./assets/swiftproxy.png" alt="Swiftproxy" width="150"></a></td>
 <td>Swiftproxy 提供 9000万+ 纯净住宅 IP，覆盖全球 220+ 个国家和地区，支持 HTTP(S)/SOCKS5、IP 轮换、Sticky Session 及精准地域定位。帮助 AI API 工具和自动化工作流从不同地区稳定访问在线服务，适用于 API 请求、网页访问、数据采集及地域测试等场景。住宅代理低至 &#36;0.7/GB，支持免费测试，使用优惠码 PROXY90 可享 9 折优惠。<a href="https://www.swiftproxy.net/?code=PR67S9A95">立即体验 Swiftproxy</a>。</td>
 </tr>
@@ -204,14 +200,6 @@ CLI 封装器，用于通过 CLIProxyAPI OAuth 即时切换多个 Claude 账户
 
 原生 Windows CLIProxyAPI 分支，集成 TUI、系统托盘及多服务商 OAuth 认证，专为 AI 编程工具打造，无需 API 密钥。
 
-### [Claude Proxy VSCode](https://github.com/uzhao/claude-proxy-vscode)
-
-一款 VSCode 扩展，提供了在 VSCode 中快速切换 Claude Code 模型的功能，内置 CLIProxyAPI 作为其后端，支持后台自动启动和关闭。
-
-### [ZeroLimit](https://github.com/0xtbug/zero-limit)
-
-Windows 桌面应用，基于 Tauri + React 构建，用于通过 CLIProxyAPI 监控 AI 编程助手配额。支持跨 Gemini、Claude、OpenAI Codex 和 Antigravity 账户的使用量追踪，提供实时仪表盘、系统托盘集成和一键代理控制，无需 API 密钥。
-
 ### [CPA-XXX Panel](https://github.com/ferretgeek/CPA-X)
 
 面向 CLIProxyAPI 的 Web 管理面板，提供健康检查、资源监控、日志查看、自动更新、请求统计与定价展示，支持一键安装与 systemd 服务。

diff --git a/README_JA.md b/README_JA.md
@@ -80,10 +80,6 @@ PackyCodeは当ソフトウェアのユーザーに特別割引を提供して
 <td>Cubenceのスポンサーシップに感謝します！Cubenceは信頼性が高く効率的なAPIリレーサービスプロバイダーで、Claude Code、Codex、Geminiなどのリレーサービスを提供しています。Cubenceは当ソフトウェアのユーザーに特別割引を提供しています：<a href="https://cubence.com/signup?code=CLIPROXYAPI&source=cpa">こちらのリンク</a>から登録し、チャージ時にプロモーションコード「CLIPROXYAPI」を入力すると10%割引になります。</td>
 </tr>
 <tr>
-<td width="180"><a href="https://go.apimart.ai/gh-cliproxyapi"><img src="./assets/apimart-en.png" alt="APIMart" width="150"></a></td>
-<td>APIMartによる本プロジェクトへのご支援に感謝します！APIMartは、AI画像・動画生成に特化した低価格APIプラットフォームです。GPT-Image-2は1枚あたりわずか&#36;0.006で、1ドルで160枚以上の画像を生成できます。画像と動画の両方を1つの非同期APIで扱えます。タスクを送信してIDを取得し、ポーリングまたはコールバックで結果を受け取れます。数万枚規模の画像をタイムアウトなしでバッチ生成でき、コードを変更せずにモデルを切り替えられます。従量課金制で月額料金は不要です。<a href="https://go.apimart.ai/gh-cliproxyapi">こちらの登録リンク</a>からすぐに始められます。</td>
-</tr>
-<tr>
 <td width="180"><a href="https://www.swiftproxy.net/?code=PR67S9A95"><img src="./assets/swiftproxy.png" alt="Swiftproxy" width="150"></a></td>
 <td>Swiftproxyは、世界220以上の国と地域をカバーする9,000万以上のクリーンな住宅IPを提供し、HTTP(S)/SOCKS5、IPローテーション、Sticky Session、詳細な地域指定に対応しています。AI APIツールや自動化ワークフローが異なる地域からオンラインサービスへ安定してアクセスできるよう支援し、APIリクエスト、Webアクセス、データ収集、地域別テストなどに最適です。住宅プロキシは&#36;0.7/GBから利用でき、無料テストにも対応しています。割引コードPROXY90の使用で10%割引になります。<a href="https://www.swiftproxy.net/?code=PR67S9A95">今すぐSwiftproxyを試す</a></td>
 </tr>
@@ -203,14 +199,6 @@ Claude、Gemini、OpenAI、Antigravityのサブスクリプションを統合し
 
 TUI、システムトレイ、マルチプロバイダーOAuthを備えたWindows向けCLIProxyAPIフォーク - AIコーディングツール用、APIキー不要
 
-### [Claude Proxy VSCode](https://github.com/uzhao/claude-proxy-vscode)
-
-Claude Codeモデルを素早く切り替えるVSCode拡張機能。バックエンドとしてCLIProxyAPIを統合し、バックグラウンドでの自動ライフサイクル管理を搭載
-
-### [ZeroLimit](https://github.com/0xtbug/zero-limit)
-
-CLIProxyAPIを使用してAIコーディングアシスタントのクォータを監視するTauri + React製のWindowsデスクトップアプリ。Gemini、Claude、OpenAI Codex、Antigravityアカウントの使用量をリアルタイムダッシュボード、システムトレイ統合、ワンクリックプロキシコントロールで追跡 - APIキー不要
-
 ### [CPA-XXX Panel](https://github.com/ferretgeek/CPA-X)
 
 CLIProxyAPI向けの軽量Web管理パネル。ヘルスチェック、リソース監視、リアルタイムログ、自動更新、リクエスト統計、料金表示機能を搭載。ワンクリックインストールとsystemdサービスに対応

diff --git a/assets/apimart-en.png b/assets/apimart-en.png
diff --git a/assets/apimart-zh.png b/assets/apimart-zh.png
diff --git a/cmd/fetch_devin_models/main.go b/cmd/fetch_devin_models/main.go
@@ -429,6 +429,15 @@ var devinCompoundSuffixes = []struct {
 	{suffix: "-thinking", effort: ""},
 	{suffix: "-max-1m", effort: "max", readd: "-1m"},
 	{suffix: "-none-1m", effort: "none", readd: "-1m"},
+	// Lightning tier keeps the effort before the -lightning suffix upstream
+	// (e.g. swe-2-high-lightning), so it aggregates as <base>-lightning.
+	{suffix: "-minimal-lightning", effort: "minimal", readd: "-lightning"},
+	{suffix: "-low-lightning", effort: "low", readd: "-lightning"},
+	{suffix: "-medium-lightning", effort: "medium", readd: "-lightning"},
+	{suffix: "-high-lightning", effort: "high", readd: "-lightning"},
+	{suffix: "-xhigh-lightning", effort: "xhigh", readd: "-lightning"},
+	{suffix: "-max-lightning", effort: "max", readd: "-lightning"},
+	{suffix: "-none-lightning", effort: "none", readd: "-lightning"},
 }
 
 var devinSimpleEffortSuffixes = []struct {
@@ -571,12 +580,19 @@ func aggregateModels(raw []rawDevinModel) []devinModelJSON {
 			}
 		}
 
+		// Lightning display names strip both the effort word and "Lightning";
+		// re-append it so swe-2-lightning shows as "SWE-2 Lightning".
+		displayName := entry.displayName
+		if strings.HasSuffix(id, "-lightning") && !strings.HasSuffix(strings.ToLower(displayName), "lightning") {
+			displayName = strings.TrimSpace(displayName + " Lightning")
+		}
+
 		res = append(res, devinModelJSON{
 			ID:                        entry.baseID,
 			Object:                    "model",
 			Type:                      "devin",
 			OwnedBy:                   vendorName(entry.vendorID, entry.baseID),
-			DisplayName:               entry.displayName,
+			DisplayName:               displayName,
 			ContextLength:             entry.contextLength,
 			MaxCompletionTokens:       64000,
 			SupportedInputModalities:  modalities,
@@ -595,7 +611,7 @@ var devinDisplayNameSuffixes = []string{
 	" Low Thinking", " Medium Thinking", " High Thinking", " XHigh Thinking",
 	" Max Thinking", " No Thinking",
 	" Low", " Medium", " High", " XHigh", " Max", " None", " Minimal",
-	" Thinking", " Fast",
+	" Thinking", " Fast", " Lightning",
 }
 
 func cleanDisplayName(label string) string {

diff --git a/cmd/fetch_devin_models/main_test.go b/cmd/fetch_devin_models/main_test.go
@@ -24,6 +24,9 @@ func TestSplitDevinUID(t *testing.T) {
 		{"swe-1-6-slow", "swe-1-6-slow", ""},
 		{"swe-1-6-fast", "swe-1-6", ""},
 		{"swe-2", "swe-2", ""},
+		{"swe-2-medium-lightning", "swe-2-lightning", "medium"},
+		{"swe-2-high-lightning", "swe-2-lightning", "high"},
+		{"swe-2-max-lightning", "swe-2-lightning", "max"},
 	}
 
 	for _, tt := range tests {

diff --git a/config.example.yaml b/config.example.yaml
@@ -34,6 +34,15 @@ server:
   # Default is 0 (disabled / no limit).
   max-request-body-bytes: 0
 
+  # Negotiated compression of downstream HTTP responses (including SSE streams).
+  # Responses are compressed only when the client advertises zstd or gzip in
+  # Accept-Encoding; WebSocket upgrades are never wrapped.
+  response-compression:
+    # Default: true.
+    enabled: true
+    # Non-streaming responses smaller than this many bytes are sent uncompressed.
+    min-bytes: 1024
+
   # Global token for GitHub requests (for example, release and asset downloads).
   # A non-empty config token takes precedence over the GITHUB_TOKEN environment variable.
   # Leave empty to use GITHUB_TOKEN, or unauthenticated requests if neither is set.
@@ -58,6 +67,9 @@ server:
 
   # Enable local network mDNS / DNS-SD service discovery (RFC 6762 / RFC 6763).
   # When enabled, broadcasts _ai-gateway._tcp with API protocol subtypes on the local network.
+  # TXT records follow draft-su-ai-gateway-discovery-00:
+  #   txtvers, path, tls, auth, protocols, auth_type, product, id, role
+  # The TXT protocols list mirrors the advertised subtypes. Management endpoints are never advertised.
   # NOTE for Docker users: Link-local multicast (224.0.0.251) cannot cross default NAT bridge networks.
   # To broadcast across the physical LAN, run the container with host networking (--net=host / network_mode: host).
   discovery:
@@ -70,12 +82,12 @@ server:
     #   - "_messages"
     #   - "_generate-content"
     #   - "_interactions"
+    #   # Optional extended subtypes: _completions, _responses-ws, _realtime, _images,
+    #   # _videos, _audio, _count-tokens, _search
     # interfaces:
     #   include: []                     # White-listed interfaces (empty: auto-detect physical interfaces)
     #   exclude: []                     # Additional interface patterns to exclude
-    # auth-required: true               # Advertise that credentials are required
-
-    # advertise-management: false      # Advertise management availability
+    # auth-required: true               # Advertise auth=1 (credentials required)
 
 # Management API settings
 management:
@@ -342,6 +354,17 @@ upstream:
     # Experimental full-duplex Codex Responses WebSocket steering (default: false).
     # Keeps one upstream account/model per socket; accepted input is never replayed.
     response-steering: false
+    # Compress request bodies sent to the ChatGPT Codex backend with zstd,
+    # mirroring the official Codex client. Applies only to OAuth credentials
+    # targeting https://chatgpt.com/backend-api/codex (API-key credentials,
+    # custom base URLs, and SIWC requests are never compressed). A 400/415
+    # rejection is retried once with an uncompressed body. Default: false.
+    upstream-request-compression: false
+    # Request bodies smaller than this many bytes are sent uncompressed.
+    upstream-request-compression-min-bytes: 1024
+    # Enable permessage-deflate compression on Codex WebSocket connections
+    # (upstream and downstream) after negotiation. Default: true.
+    websocket-compression: true
     # Disable forcing the official Codex User-Agent and Originator headers on HTTP/SSE and WebSocket requests.
     disable-codex-cloaking: false
     # Hold back the frames that carry nothing the client has seen - the handshake (response.created,

diff --git a/docs/lan-discovery.md b/docs/lan-discovery.md
@@ -0,0 +1,86 @@
+# LAN AI gateway discovery
+
+CPA advertises `_ai-gateway._tcp` using the profile in
+[draft-su-ai-gateway-discovery-00](https://datatracker.ietf.org/doc/draft-su-ai-gateway-discovery/).
+Discovery is opt-in. Management routes are never advertised; the deprecated
+`advertise-management` configuration field is accepted but ignored.
+
+## TXT records
+
+Advertisements place `txtvers`, `path`, `tls`, and `auth` first. Optional keys
+are `protocols`, `auth_type`, `product`, `id`, and `role`, in that order.
+Each character-string is at most 255 bytes and the complete TXT payload,
+including length prefixes, is at most 400 bytes. Optional values are omitted
+whole when they do not fit. The capability list is derived from advertised
+subtypes. Version 1 booleans use `0` and `1`.
+
+Browsers compare ASCII keys case-insensitively without trimming whitespace.
+The first occurrence wins, including bare keys. Unsupported `txtvers` or invalid
+required attributes cannot fall back to the old CPA schema. Known draft values
+must satisfy the profile's printable ASCII and token syntax. Unknown protocol
+and authentication tokens remain in `raw_txt` but are not exposed as recognized
+capabilities or assigned operation URLs.
+
+Legacy CPA advertisements using `version=1`, `api_*`, and `auth_required` are
+accepted separately. If their optional `protocols` list was omitted, only the
+advertised API base paths are exposed; no operation support is inferred.
+
+## Resolution and interface scope
+
+The browser uses a bounded, per-interface DNS record cache rather than the
+zeroconf dependency's single-packet browse implementation. PTR, SRV, TXT, A,
+and AAAA may arrive in separate packets in any order. Missing records are
+queried with exponential backoff, while known PTR lifetimes independently
+trigger refreshes before expiry. Repeated queries include fresh known answers;
+long lists use bounded, TC-marked continuation packets. TTL expiration,
+cache-flush, and goodbye one-second grace periods apply. At cache capacity,
+flush replacements take precedence over the old records being retired. Expired records do not remain in returned
+results. Full subtype queries such as `_responses._sub._ai-gateway._tcp` resolve
+the PTR target's main service records.
+
+Only responses from UDP port 5353 on selected interfaces are accepted. Receive
+interface and destination control messages are required; platforms unable to
+provide them fail closed. Multicast responses must target the standard mDNS
+group; unicast responses must target a local interface address and originate
+on-link (RFC 6762 Section 11). Joining a multicast group alone is not treated
+as receive isolation. Default interface selection follows CPA's physical LAN
+filter and never falls back to all interfaces.
+
+Loopback, unspecified, multicast, and limited-broadcast addresses are filtered.
+IPv6 link-local results retain their receiving interface in `ipv6_zones`, keyed
+by address string. `interface_index` distinguishes same-name instances on
+different links. Consumers must use the zone when connecting to a link-local
+address. CLI WebSocket URLs use `ws`/`wss`; scoped URI hosts escape the zone
+separator as `%25`. Unscoped link-local addresses are not offered as usable
+connection targets.
+
+The cache admits at most 256 service instances, 17,408 resource records, and
+4 MiB of accounted record storage. Address RRsets are capped at 64 combined
+IPv4/IPv6 addresses per host/interface; TXT records have at most 64 strings.
+Legacy TXT input is capped at 16 KiB; draft input remains capped at 400 bytes.
+These bounds constrain retained data, not transient allocations made by the
+DNS packet decoder or exact process heap usage.
+
+`Browse` honors its caller's context without imposing an internal deadline.
+Cancellation closes sockets, joins reader goroutines, and returns the currently
+valid partial results. Reaching 256 resolved services may end the scan early.
+
+## Trust boundary
+
+mDNS is unauthenticated. Source/interface validation reduces accidental and
+off-link candidate pollution; it does not authenticate an on-link advertiser.
+Neither `product`, `id`, instance names, nor advertised hostnames establish trust.
+The discover CLI displays candidates and never connects to them or sends keys.
+Clients must obtain user approval or match preconfigured trust rules before
+connecting, validate TLS identity independently when applicable, and never
+forward credentials to a different origin automatically. Querying model
+catalogs is optional; `auth=1` catalogs require credentials.
+
+## Verification
+
+Deterministic tests exercise all 120 orderings of the five resolution records,
+TTL/cache-flush/goodbye behavior, per-interface separation, source acceptance,
+subtype resolution, size boundaries, legacy compatibility, and scoped URLs.
+The existing live multicast integration test additionally checks discovery of
+CPA's advertiser on the local machine. This is not a complete RFC conformance
+or cross-platform network test suite.