omnilane 0.31.0 → 0.33.0

package/README.md

@@ -78,11 +78,11 @@ key with no extra CLI at all):
 ```mermaid
 flowchart LR
     M["main loop<br/><i>any CLI you drive</i>"] --> T{{"routing.yaml<br/>one shared table"}}
-    T -->|hardest-coding| C1["Codex — GPT-5.6 Sol"]
-    T -->|bulk-mechanical| C2["Codex — GPT-5.6 Terra"]
-  T -->|taste-final| C3["Claude — Opus 5"]
-    T -->|long-context| C4["Gemini — 3.1 Pro"]
-    T -->|live-search| C5["Grok — 4.5"]
+    T -->|hardest-coding| C1["Claude — Fable 5.1"]
+    T -->|bulk-mechanical| C2["Codex — GPT-5.6 Sol"]
+    T -->|taste-final| C3["Claude — Fable 5.1"]
+    T -->|long-context| C4["Gemini — 3.7 Flash"]
+    T -->|live-search| C5["Grok — 4.6"]
     T -->|"arbitrate (opt-in)"| C6["vote — 1-4 model panel"]
 ```
 
@@ -115,23 +115,25 @@ actually resolves.
 
 | Lane | First choice | Backup | When |
 |---|---|---|---|
-| 🔥 hardest-coding | GPT-5.6 Sol (xhigh) | Claude Opus 5 (xhigh) | Hardest implementation, deep root-cause debug, correctness-critical edits |
-| 🏗️ bulk-mechanical | GPT-5.6 Terra (max) | Claude Sonnet 5 (high) | Refactors, migrations, tests, review sweeps — mechanical endurance |
-| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.6 Flash (Low) | High-volume scans, first-pass filtering |
-| ⚖️ hard-judgment | Claude Opus 5 (xhigh) | GPT-5.6 Sol (max) | Architecture arbitration, deep reasoning, second opinions |
-| ✒️ taste-final | Claude Opus 5 (high) | GPT-5.6 Sol (max) | User-facing prose, prompt/doc polish, style arbitration |
-| 💬 consult | Explicit named vendor/model | — (no fallback) | Direct natural-language consultation; always keep `--vendor` |
-| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 5 (high) | UI drafts only WITH a design system / reference images |
-| 📚 long-context | Gemini 3.1 Pro (High) | GPT-5.6 Sol (high) | 1M-token sweeps, retrieval, and synthesis across long documents; Flash for fast repeated loops |
-| ⚡ fast-agentic | GPT-5.6 Luna (max) | Gemini 3.6 Flash (High) | Fast multi-step agentic loops, multimodal checks |
-| 📡 live-search | Grok 4.5 | — (off) | Realtime X/web search and social context |
-| 🚰 coding-overflow | Grok 4.5 | Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex-quota relief valve for mid-tier coding |
+| 🔥 hardest-coding | Claude Fable 5.1 (xhigh) | GPT-5.6 Sol (xhigh) | Hardest implementation, deep root-cause debug, correctness-critical edits |
+| 🏗️ bulk-mechanical | GPT-5.6 Sol (high) | Gemini 3.7 Flash (High) → Claude Sonnet 5 (high) | Refactors, migrations, tests, review sweeps — mechanical endurance |
+| 🧹 triage | GPT-5.6 Luna (high) | Gemini 3.7 Flash (Low) → Claude Haiku 4.5 | High-volume scans, first-pass filtering |
+| ⚖️ hard-judgment | Claude Fable 5.1 (xhigh) | GPT-5.6 Sol (max) → Grok 4.6 | Architecture arbitration, deep reasoning, second opinions |
+| ✒️ taste-final | Claude Fable 5.1 (high) | GPT-5.6 Sol (max) | User-facing prose, prompt/doc polish, style arbitration |
+| 💬 consult | GPT-5.6 Sol (max) | Claude Fable 5.1 (high) → Grok 4.6 → Gemini 3.7 Flash (High) | Direct named-model consultation; keep `--vendor` to prevent fallback |
+| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Fable 5.1 (high) | UI drafts only WITH a design system / reference images |
+| 📚 long-context | Gemini 3.7 Flash (Medium) | GPT-5.6 Terra (max) → Claude Opus 5 (medium) | Long-document retrieval and synthesis, ordered on AA-LCR, cost, and throughput |
+| ⚡ fast-agentic | Gemini 3.7 Flash (Medium) | GPT-5.6 Luna (high) | Fast multi-step agentic loops, multimodal checks |
+| 📡 live-search | Grok 4.6 | — (off) | Realtime X/web search and social context |
+| 🚰 coding-overflow | Grok 4.6 | Gemini 3.7 Flash (High) → Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex-quota relief valve for mid-tier coding |
 | 🗳️ arbitrate | off (opt-in vote panel) | — | Built-in opinion panel for big calls — disabled by default; enable it in `routing.local.yaml`, one call per voter per round |
 
 The **backup** is the next candidate in the lane's `routing.yaml` chain — what
 dispatch falls back to when the first-choice vendor CLI is not installed. Every
 lane is such a chain; when nothing in it is installed the lane degrades to `off`.
 
+> **Fable 5.1 is in the defaults — and where Opus 5 still fits.** See the current three-way evidence and Opus override in the [FAQ](#-faq).
+
 ### Natural-language consultation
 
 With the `omnilane` skill or `/route`, you can ask normally:
@@ -155,12 +157,12 @@ with who is driving. What changes is which lanes you **self-execute** (you
 already are that model, so no second call) versus **dispatch**. Your harness's
 `omnilane` skill applies the right row automatically; this is the human view.
 
-- **Claude Code · Fable 5** — self-execute: hard-judgment, taste-final, the hardest correctness-critical fixes. Dispatch mechanical coding volume → Codex, long-context → Gemini, live-search → Grok.
-- **Claude Code · Opus 5** — self-execute: hard-judgment and taste-final. Dispatch bulk coding to Codex lanes, long-context → Gemini, live-search → Grok.
-- **Codex · Sol** — self-execute: hardest-coding, hard-judgment, ui-draft. Dispatch taste-final → Claude, long-context → Gemini, live-search → Grok, bulk → Codex Terra.
-- **Codex · Terra** — self-execute: bulk-mechanical. Escalate the genuinely hardest pieces to Sol; dispatch taste → Claude, long-context → Gemini, live-search → Grok.
-- **Grok Build · Grok 4.5** — self-execute: live-search, coding-overflow (mid-tier coding). Dispatch everything hard to Codex/Claude/Gemini — and verify every API signature and cited fact first.
-- **Antigravity · Gemini** — self-execute: long-context and context-heavy agentic work on 3.1 Pro, fast repeated loops on Flash. Dispatch hardest coding/judgment/taste to Codex/Claude; live-search → Grok.
+- **Claude Code · Fable 5.1** — self-execute: hard-judgment, taste-final, hardest-coding. Dispatch bulk → Codex Sol high; long-context and fast loops → Gemini 3.7 Flash; live-search → Grok.
+- **Claude Code · Opus 5** — self-execute: hard-judgment and taste-final when its lower hallucination rate or price is preferred. Dispatch hardest coding → Fable 5.1 or Sol, bulk → Sol high, long-context and fast loops → Gemini 3.7 Flash, live-search → Grok.
+- **Codex · Sol** — self-execute: hardest-coding, bulk-mechanical, hard-judgment, ui-draft. Dispatch taste-final → Claude, long-context and fast loops → Gemini 3.7 Flash, live-search → Grok.
+- **Codex · Terra** — self-execute: long-context as the Codex fallback. Bulk-mechanical now defaults to Sol high; escalate hardest pieces to Sol xhigh, taste → Claude, fast loops → Gemini 3.7 Flash, live-search → Grok.
+- **Grok Build · Grok 4.6** — self-execute: live-search and coding-overflow. Dispatch hard coding/judgment/taste to Codex/Claude/Gemini; verify API signatures and cited facts.
+- **Antigravity · Gemini 3.7 Flash** — self-execute: long-context and fast loops at Medium, bulk/overflow at High, triage at Low. Dispatch hardest coding/judgment/taste to Codex/Claude; live-search → Grok.
 
 </details>
 
@@ -319,7 +321,7 @@ omnilane ui url                                # print the current authenticated
 omnilane ui stop                               # stop the Live UI
 omnilane doctor [--json] [--strict] [--probe V] [--probe-timeout SEC]  # live probe is opt-in
 omnilane benchmark [--json] [--run] [--vendor V] [--cost-per-call V=USD] # dry-run by default
-dispatch.sh [--background] [--dry-run] [--mode advise|work|sysops] [--workdir DIR]
+dispatch.sh [--background] [--dry-run] [--thread NAME] [--mode advise|work|sysops] [--workdir DIR]
             [--vendor V] [--model M] [--effort E] [--timeout SEC] [--job-timeout SEC]
             LANE "TASK"                              # "-" reads task from stdin
 dispatch.sh [--json] --list [--json]
@@ -338,6 +340,11 @@ configure.sh                                        # interactive lane menu
 configure.sh set|get|unset|list|diff LANE [SPEC]    # script/inspect routing.local.yaml, no tty
 ```
 
+`--thread NAME` continues named Claude, Codex, Grok, or Gemini conversations across
+single-shot dispatches. In 0.33.0 it pins vendor, model, effort, and physical
+workdir; use `jobs.sh threads`, `threads show NAME`, or `threads rm NAME` to
+manage local state without deleting the vendor session.
+
 `jobs recommend` reads only validated public metadata and exit codes. It ranks
 eligible vendors by success rate, sample count, then name; the default minimum
 is three completed jobs. It never reads task/result bodies or changes routing.
@@ -469,7 +476,7 @@ whose CLI is actually installed. With one subscription the whole table collapses
 onto that vendor; lanes with nothing available turn off rather than failing.
 `omnilane doctor` shows exactly what your machine can reach today, and
 `routing.local.yaml.example` ships starter profiles for common situations
-(Claude-only, Codex-heavy, no-Codex).
+(Claude-focused, Codex-heavy, no-Codex).
 
 </details>
 
@@ -489,52 +496,30 @@ with the key you set — those are advise-only and never edit files.
 </details>
 
 <details>
-<summary><b>Where is Claude Fable 5? Why is it not in the default table?</b></summary>
+<summary><b>Fable 5.1 is in the defaults — and where Opus 5 still fits</b></summary>
 
 <br/>
 
-**Because the top Claude tier is usually the main loop itself, not a dispatched
-worker.** Lanes exist to send work to a model *other than* the one you are
-driving. If Fable 5 is your main loop, routing judgment and taste back to Fable 5
-just adds a second call for no gain — which is why the "pick your main model"
-list above gives Fable 5 its own row as a **driver**, self-executing
-hard-judgment, taste-final, and the hardest correctness-critical fixes.
-
-**The measurements do not argue for it as a worker either.** On the Artificial
-Analysis Intelligence Index (2026-07-24) Opus 5 (max) scores 61 and Fable 5 (max)
-scores 60 — Artificial Analysis calls them "effectively tied", and Epoch AI's
-Capability Index ranks them the other way (Fable 5 161, Opus 5 159). Call it a
-draw on general intelligence. Where they are not tied is agentic professional
-output, and Opus 5 leads by a wide margin:
-
-| Benchmark | Claude Opus 5 (max) | Claude Fable 5 | |
-|---|---:|---:|---|
-| AA-Briefcase (agentic knowledge work, Elo) | 1720 | 1574 | **+146** |
-| GDPval-AA v2 (Elo) | 1861 | 1747 | **+114** |
-| Cost per AA-Briefcase task | $17.79 | $22.30 | **-20%** |
-| API price, input / output per 1M | $5 / $25 | $10 / $50 | **half** |
-
-Opus 5's max, xhigh and high tiers sweep the top three AA-Briefcase places, and
-its `high` tier still beats Fable 5 at under half the cost per task. So Fable 5
-costs twice as much without buying an advantage on any axis a lane is defined
-around.
-
-**What Fable 5 is genuinely better at**: factual breadth. It stays ahead of
-Opus 5 on AA-Omniscience, as its size class suggests, and Opus 5 answers more
-readily when uncertain — its hallucination rate is 50%, up 14 points from
-Opus 4.8. If your task is recall-heavy rather than execution-heavy, name
-Fable 5 explicitly:
+Fable 5.1 now leads `hardest-coding`, `hard-judgment`, and `taste-final`.
+At matched xhigh effort it leads Opus 5 on intelligence, agentic work, and
+coding. Sol max remains the far cheaper cross-vendor judgment fallback.
 
-```bash
-dispatch.sh --vendor claude --model claude-fable-5 --effort high consult "…"
-```
+| Benchmark (AA, retrieved 2026-09-02) | Claude Fable 5.1 (xhigh) | Claude Opus 5 (xhigh) | GPT-5.6 Sol (max) |
+|---|---:|---:|---:|
+| Intelligence | 64.8 | 62.5 | 60.9 |
+| Agentic | 59.8 | 58.4 | 57.8 |
+| Coding | 80.7 | 77.0 | 77.4 |
+| Hallucination rate (lower is better) | .71 | **.60** | .92 |
+| AA $/task | $2.65 | $1.80 | **$0.95** |
 
-**This is a cost / main-loop policy choice, not a capability verdict.** Fable 5
-is in the configurator's model menu, and one line in `routing.local.yaml`
-overrides the default if you disagree:
+Fable 5.1 is not a bulk or triage default: it costs twice Opus 5 per token and
+consumes the most Claude Code subscription quota per turn. Opus 5 remains the
+lower-hallucination, lower-price Claude option, stays in `long-context` at
+medium, and remains selectable everywhere through
+`~/.omnilane/routing.local.yaml`:
 
 ```yaml
-taste-final: claude claude-fable-5 high
+hard-judgment: claude claude-opus-5 xhigh
 ```
 
 </details>
@@ -598,7 +583,7 @@ Default lane assignments follow Artificial Analysis coding/intelligence data
 pages) plus published head-to-head reviews; they are opinions, not laws — the
 configurator and `routing.local.yaml` exist so you can disagree. The full
 working notes, including per-benchmark caveats, live in
-[`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
+[`docs/model-capabilities-2026-09.md`](docs/model-capabilities-2026-09.md).
 
 ## ⚠️ Known limitations
 
@@ -614,6 +599,21 @@ working notes, including per-benchmark caveats, live in
 
 ## 📜 Release history
 
+## What's new in v0.33.0
+
+- **Four-vendor threaded dispatch.** `--thread NAME` continues pinned Claude,
+  Codex, Grok, or Gemini conversations across foreground or background
+  single-shot jobs; direct-API vendors, `exec`, live mode, and pin mismatches
+  stop with visible exit-2 notices.
+- **Thread inspection.** `jobs.sh threads`, `threads show NAME`, and
+  `threads rm NAME` list, inspect, or remove local thread state.
+
+## What's new in v0.32.0
+
+- **Full 2026-09 routing re-evaluation.** Fable 5.1 and Gemini 3.7 Flash enter the defaults, backed by the dated Artificial Analysis snapshot.
+- **Catalogs match live CLI surfaces.** Fable 5.1 is added and retired Gemini 3.5 Flash rows are removed.
+- **Opus 5 remains available.** It stays in `long-context` and can override any lane through `routing.local.yaml`.
+
 ## What's new in v0.31.0
 
 - **Unlimited goal budgets by default.** `budget_jobs` and `budget_seconds` now persist as JSON `null` and render as `unlimited`; the previous implicit 8-job and 900-second caps are gone. Use `--budget-jobs N` or `--budget-seconds S` to opt in to a hard cap. The repeat-failure fuse remains enabled by default.
@@ -715,7 +715,7 @@ working notes, including per-benchmark caveats, live in
   latency-bound.
 - **Lane comments no longer carry numbers.** `routing.yaml` now states why each
   ordering holds; every score, price and throughput figure lives in
-  `docs/model-capabilities-2026-07.md` with its retrieval date, so a stale
+  `docs/model-capabilities-2026-09.md` with its retrieval date, so a stale
   figure never requires a routing-table edit.
 - **A value profile** in `routing.local.yaml.example` trades about one
   Intelligence Index point for 30-40% off the cost per task.
@@ -774,7 +774,7 @@ working notes, including per-benchmark caveats, live in
   profiles still pointed at `claude-opus-4-8` and Gemini 3.5 Flash; they now use
   Claude Opus 5 (with lane-appropriate effort) and Gemini 3.6 Flash.
 - **Corrected the Intelligence Index figures** in
-  `docs/model-capabilities-2026-07.md` against the Artificial Analysis source
+  `docs/model-capabilities-2026-09.md` against the Artificial Analysis source
   (index points, not percentages), added the AA-Briefcase / GDPval-AA v2
   comparison, and recorded the two results that cut against the defaults:
   Fable 5's lead on factual knowledge and Sol's lead on presentation quality.
@@ -822,7 +822,7 @@ working notes, including per-benchmark caveats, live in
 - **Five OpenAI-compatible direct-API vendors** — `deepseek`, `zai` (GLM),
   `mistral`, `groq`, and `cerebras` join `openrouter` as CLI-free lanes (curl +
   a `<VENDOR>_API_KEY`). A one-line `lib/common.sh` registry entry adds each;
-  see [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
+  see [`docs/model-capabilities-2026-09.md`](docs/model-capabilities-2026-09.md).
 - **Fish shell completion** — `omnilane completion fish | source`.
 
 ## What's new in v0.8.3
@@ -846,7 +846,7 @@ working notes, including per-benchmark caveats, live in
   DeepSeek, Z.ai GLM, Mistral, Groq, and Cerebras. Each needs only `curl` and
   its `<VENDOR>_API_KEY`; advise/consult only. A one-line `lib/common.sh`
   registry entry defines each endpoint, key env, and default model. See
-  [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
+  [`docs/model-capabilities-2026-09.md`](docs/model-capabilities-2026-09.md).
 - **`opencode` vendor** — headless dispatch through the OpenCode
   multi-provider aggregator CLI (`opencode run`). Advise mode pins OpenCode's
   built-in read-only `plan` agent; work mode uses `--auto`. Joins the default

package/README.zh-CN.md

@@ -72,11 +72,11 @@ omnilane 让**任何**一个 agentic CLI 的主循环把子任务分类到通道
 ```mermaid
 flowchart LR
     M["主循环<br/><i>你在用的任一 CLI</i>"] --> T{{"routing.yaml<br/>一张共用路由表"}}
-    T -->|hardest-coding| C1["Codex — GPT-5.6 Sol"]
-    T -->|bulk-mechanical| C2["Codex — GPT-5.6 Terra"]
-    T -->|taste-final| C3["Claude — Opus 5"]
-    T -->|long-context| C4["Gemini — 3.1 Pro"]
-    T -->|live-search| C5["Grok — 4.5"]
+    T -->|hardest-coding| C1["Claude — Fable 5.1"]
+    T -->|bulk-mechanical| C2["Codex — GPT-5.6 Sol"]
+    T -->|taste-final| C3["Claude — Fable 5.1"]
+    T -->|long-context| C4["Gemini — 3.7 Flash"]
+    T -->|live-search| C5["Grok — 4.6"]
     T -->|"arbitrate(可选)"| C6["vote — 1-4 模型评审团"]
 ```
 
@@ -103,24 +103,23 @@ flowchart LR
 
 | 通道 | 首选模型 | 备选模型 | 用途 |
 |---|---|---|---|
-| 🔥 hardest-coding | GPT-5.6 Sol (xhigh) | Claude Opus 5 (xhigh) | 最难的实现、深度调试、正确性攸关的修改 |
-| 🏗️ bulk-mechanical | GPT-5.6 Terra (max) | Claude Sonnet 5 (high) | 重构、迁移、测试、大面积扫描——机械耐力活 |
-| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.6 Flash (Low) | 高量初筛、第一轮过滤 |
-| ⚖️ hard-judgment | Claude Opus 5 (xhigh) | GPT-5.6 Sol (max) | 架构仲裁、深度推理、第二意见 |
-| ✒️ taste-final | Claude Opus 5 (high) | GPT-5.6 Sol (max) | 对外文字、prompt 与文档打磨、风格终审 |
-| 💬 consult | 明确指定的厂商/模型 | —(不降级) | 自然语言直接咨询;必须保留 `--vendor` |
-| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 5 (high) | 有设计规范/参考图时的 UI 出稿;开放式视觉品味交给 taste-final |
-| 📚 long-context | Gemini 3.1 Pro (High) | GPT-5.6 Sol (high) | 百万 token 扫读、检索与长文跨段整合;高速重复循环仍优先 Flash |
-| ⚡ fast-agentic | GPT-5.6 Luna (max) | Gemini 3.6 Flash (High) | 快速多步骤 agentic 循环、多模态检查 |
-| 📡 live-search | Grok 4.5 | —(off) | 实时 X/网络搜索与社群脉络 |
-| 🚰 coding-overflow | Grok 4.5 | Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex 额度吃紧时的中量级编码溢流道;事实性声明须另行查证 |
-| 🗳️ arbitrate | off(可选评审团) | — | 内置意见评审团,重大决定用——默认关闭,要用在 `routing.local.yaml` 打开;每评审每轮烧一次额度 |
+| 🔥 hardest-coding | Claude Fable 5.1 (xhigh) | GPT-5.6 Sol (xhigh) | 最难的实现、深度调试、正确性关键的修改 |
+| 🏗️ bulk-mechanical | GPT-5.6 Sol (high) | Gemini 3.7 Flash (High) → Claude Sonnet 5 (high) | 重构、迁移、测试、大范围扫描——机械耐力活 |
+| 🧹 triage | GPT-5.6 Luna (high) | Gemini 3.7 Flash (Low) → Claude Haiku 4.5 | 大量扫描、第一轮筛选 |
+| ⚖️ hard-judgment | Claude Fable 5.1 (xhigh) | GPT-5.6 Sol (max) → Grok 4.6 | 架构裁决、深度推理、第二意见 |
+| ✒️ taste-final | Claude Fable 5.1 (high) | GPT-5.6 Sol (max) | 对外文字、提示词／文档润色、风格裁决 |
+| 💬 consult | GPT-5.6 Sol (max) | Claude Fable 5.1 (high) → Grok 4.6 → Gemini 3.7 Flash (High) | 直接指定模型咨询；保留 `--vendor` 避免降级 |
+| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Fable 5.1 (high) | 仅在提供设计系统／参考图时生成 UI 草稿 |
+| 📚 long-context | Gemini 3.7 Flash (Medium) | GPT-5.6 Terra (max) → Claude Opus 5 (medium) | 长文档提取与综合，按 AA-LCR、成本和吞吐排序 |
+| ⚡ fast-agentic | Gemini 3.7 Flash (Medium) | GPT-5.6 Luna (high) | 高速多步骤工具循环、多模态检查 |
+| 📡 live-search | Grok 4.6 | — (`off`) | 实时 X／网页搜索与社交上下文 |
+| 🚰 coding-overflow | Grok 4.6 | Gemini 3.7 Flash (High) → Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex 配额耗尽时的中量级编码安全阀 |
+| 🗳️ arbitrate | `off`（可选模型评审团） | — | 重大决定的内置意见评审团；默认禁用，在 `routing.local.yaml` 启用，每位评审每轮调用一次 |
 
 **备选模型**是候选链的下一位——首选那家的厂商 CLI 没装时,派发就降到它。每条
 通道都是这样一条链;整条都没装时,通道自动降为 `off`。
 
-> **Claude Fable 5 去哪了?** 默认表刻意不放——理由与完整数据见
-> [常见问题](#-常见问题)。
+> **Fable 5.1 已进入默认表——以及 Opus 5 仍适合放在哪里。** 三方数据与 Opus override 见[常见问题](#-常见问题)。
 
 ### 自然语言咨询
 
@@ -141,12 +140,12 @@ flowchart LR
 你哪些通道**自己做**(你本来就是那个模型,省一次调用)、哪些**派出去**。你 CLI 里
 的 `omnilane` 技能会自动套对的那一行,这里是给人看的版本。
 
-- **Claude Code · Fable 5** — 自己做:hard-judgment、taste-final、最吃正确性的硬修。派出去:机械编码量 → Codex、长文 → Gemini、实时搜索 → Grok。
-- **Claude Code · Opus 5** — 自己做：hard-judgment、taste-final。大量编码走 Codex 通道、长文 → Gemini、实时搜索 → Grok。
-- **Codex · Sol** — 自己做:hardest-coding、hard-judgment、ui-draft。派出去:taste-final → Claude、长文 → Gemini、实时搜索 → Grok、粗活 → Codex Terra。
-- **Codex · Terra** — 自己做:bulk-mechanical。真正最硬的往上升给 Sol;taste → Claude、长文 → Gemini、实时搜索 → Grok。
-- **Grok Build · Grok 4.5** — 自己做:live-search、coding-overflow(中量级编码)。所有硬活派给 Codex/Claude/Gemini——先验每个 API 签名与引用事实。
-- **Antigravity · Gemini** — 自己做:3.1 Pro 的长文与重上下文 agentic 工作、Flash 的高速重复循环。最难的编码/判断/文字派给 Codex/Claude;实时搜索 → Grok。
+- **Claude Code · Fable 5.1**——自己执行：hard-judgment、taste-final、hardest-coding。派发：bulk → Codex Sol high；long-context／高速循环 → Gemini 3.7 Flash；实时搜索 → Grok。
+- **Claude Code · Opus 5**——需要更低幻觉率或价格时，自己执行 hard-judgment、taste-final。最难编码 → Fable 5.1 或 Sol；bulk → Sol high；long-context／高速循环 → Gemini 3.7 Flash；实时搜索 → Grok。
+- **Codex · Sol**——自己执行：hardest-coding、bulk-mechanical、hard-judgment、ui-draft。派发：taste-final → Claude；long-context／高速循环 → Gemini 3.7 Flash；实时搜索 → Grok。
+- **Codex · Terra**——自己执行 long-context 的 Codex 备用任务；bulk-mechanical 现在默认由 Sol high 处理。最难部分升级到 Sol xhigh，taste → Claude，高速循环 → Gemini 3.7 Flash，实时搜索 → Grok。
+- **Grok Build · Grok 4.6**——自己执行 live-search、coding-overflow。最难的编码／判断／文字交给 Codex、Claude、Gemini；仍需验证 API 签名和引用事实。
+- **Antigravity · Gemini 3.7 Flash**——自己执行：Medium 的 long-context／高速循环、High 的 bulk／overflow、Low 的 triage。最难编码／判断／文字交给 Codex、Claude；实时搜索 → Grok。
 
 </details>
 
@@ -266,7 +265,7 @@ omnilane ui status                             # 查看 Live UI 是否正在运
 omnilane ui url                                # 输出当前通过认证的本地链接
 omnilane ui stop                               # 停止 Live UI
 omnilane doctor [--json]                       # 只读检查路由与本地运行环境
-dispatch.sh [--background] [--dry-run] [--mode advise|work|sysops] [--workdir 目录]
+dispatch.sh [--background] [--dry-run] [--thread NAME] [--mode advise|work|sysops] [--workdir 目录]
             [--vendor V] [--model M] [--effort E] [--timeout SEC] [--job-timeout SEC]
             通道 "任务"                              # "-" 表示从 stdin 读任务
 dispatch.sh [--json] --list [--json]
@@ -284,6 +283,11 @@ configure.sh                                        # 交互通道菜单
 configure.sh set|get|unset|list|diff LANE [SPEC]    # 非交互编辑/查看 routing.local.yaml
 ```
 
+`--thread NAME` 可在多次单次派发间延续命名的 Claude、Codex、Grok 或 Gemini
+对话。0.33.0 会固定厂商、模型、effort 与实际工作目录；使用
+`jobs.sh threads`、`threads show NAME`、`threads rm NAME` 管理本地状态，
+删除状态不会删除厂商端会话。
+
 退出码:`2` 用法错误(包括厂商值无效,或指定厂商不在该通道)、`3` 通道已关闭、
 `4` 候选链没有可用 CLI,或指定厂商已配置但其 CLI 不可用、
 `5` 第一轮成功评审太少、`6` 第二轮没有任何反驳成功、`86` 拒绝嵌套派发、
@@ -395,45 +399,29 @@ omnilane goal close "$GOAL_ID" --summary "结账集成已稳定"
 </details>
 
 <details>
-<summary><b>Claude Fable 5 去哪了?为什么默认表不放它?</b></summary>
+<summary><b>Fable 5.1 已进入默认表——以及 Opus 5 仍适合放在哪里</b></summary>
 
 <br/>
 
-**因为 Claude 顶级档通常就是主循环本人,不是被派发的工人。** 通道存在的意义,
-是把工作送给「你正在开的那个模型以外」的模型。如果 Fable 5 就是你的主循环,把
-判断和文字再路由回 Fable 5 只是多一次调用、毫无增益——所以上面那份「选你的主控
-模型」清单里,Fable 5 有自己独立的一行,身份是**主控**:自己做 hard-judgment、
-taste-final、最吃正确性的硬修。
-
-**测量数据也不支持把它当工人。** Artificial Analysis 智能指数(2026-07-24)
-Opus 5(max)61 分、Fable 5(max)60 分——AA 自己的用词是「实质打平」,而 Epoch AI
-的能力指数排序还是反过来的(Fable 5 161、Opus 5 159)。综合智力就当它平手。真正
-拉开差距的是 agentic 专业产出,而且差很多:
-
-| 评测 | Claude Opus 5 (max) | Claude Fable 5 | |
-|---|---:|---:|---|
-| AA-Briefcase(agentic 知识工作,Elo) | 1720 | 1574 | **+146** |
-| GDPval-AA v2(Elo) | 1861 | 1747 | **+114** |
-| AA-Briefcase 每任务成本 | $17.79 | $22.30 | **-20%** |
-| API 单价,输入/输出 每 1M | $5 / $25 | $10 / $50 | **一半** |
+Fable 5.1 现在领跑 `hardest-coding`、`hard-judgment`、`taste-final`。
+同为 xhigh 时，它在智能、代理式工作和编码上都领先 Opus 5；Sol max 则保留为
+便宜得多的跨厂商判断备用项。
 
-Opus 5 的 max、xhigh、high 三个档位包揽 AA-Briefcase 前三名,连 `high` 档都在
-不到一半的每任务成本下赢过 Fable 5。也就是说,Fable 5 贵一倍,却换不到任何一条
-通道所在意的优势。
+| 评测（AA，抓取于 2026-09-02） | Claude Fable 5.1 (xhigh) | Claude Opus 5 (xhigh) | GPT-5.6 Sol (max) |
+|---|---:|---:|---:|
+| 智能 | 64.8 | 62.5 | 60.9 |
+| 代理式 | 59.8 | 58.4 | 57.8 |
+| 编码 | 80.7 | 77.0 | 77.4 |
+| 幻觉率（越低越好） | .71 | **.60** | .92 |
+| AA 每任务成本 | $2.65 | $1.80 | **$0.95** |
 
-**Fable 5 真正更强的地方**:事实广度。它在 AA-Omniscience 上仍领先 Opus 5
-(符合两者的规模差),而 Opus 5 在没把握时更倾向直接作答——幻觉率 50%,比
-Opus 4.8 高 14 个百分点。任务偏「回想」而非「执行」时,直接点名它:
-
-```bash
-dispatch.sh --vendor claude --model claude-fable-5 --effort high consult "…"
-```
-
-**这是成本与主循环策略的选择,不是能力判决。** 设置菜单的模型清单里有 Fable 5,
-在 `routing.local.yaml` 写一行就能覆盖默认:
+Fable 5.1 没进入 bulk 或 triage：每 token 价格是 Opus 5 的两倍，而且每轮
+消耗最多 Claude Code 订阅配额。Opus 5 仍是幻觉率更低、价格更低的 Claude
+选项，并以 medium 保留在 `long-context`；也能通过
+`~/.omnilane/routing.local.yaml` 放回任意通道：
 
 ```yaml
-taste-final: claude claude-fable-5 high
+hard-judgment: claude claude-opus-5 xhigh
 ```
 
 </details>
@@ -493,7 +481,7 @@ vendor 一律当成 `work`,而且它只能逐次明确指定,永远不是 lane
 默认关闭;要用就在 `routing.local.yaml` 写
 `arbitrate: vote codex,claude,grok -`(从四家里任选 1-4 个评审),
 或改用 `exec` 厂商指向你自己的多模型审查闸脚本。完整工作笔记(含各评测的
-但书)见 [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md)。
+但书)见 [`docs/model-capabilities-2026-09.md`](docs/model-capabilities-2026-09.md)。
 
 ## ⚠️ 已知限制
 
@@ -507,6 +495,20 @@ vendor 一律当成 `work`,而且它只能逐次明确指定,永远不是 lane
 
 ## 📜 版本历程
 
+## v0.33.0 新功能
+
+- **四厂商线程派发。** `--thread NAME` 可让固定厂商、模型、effort 与工作
+  目录的 Claude、Codex、Grok 或 Gemini 对话跨前台或后台单次作业延续；
+  direct-API 厂商、`exec`、实时模式与固定值冲突都会以清晰的退出码 2 提示停止。
+- **线程状态管理。** `jobs.sh threads`、`threads show NAME`、`threads rm NAME`
+  可列出、查看或删除本地线程状态。
+
+## v0.32.0 新功能
+
+- **基于 AA 2026-09 快照全面重评路由。** Fable 5.1 和 Gemini 3.7 Flash 进入默认表，数据集中在新的日期化文档。
+- **模型目录同步当前 CLI 接口。** 加入 Fable 5.1，移除 agy 已下架的 Gemini 3.5 Flash 项，并同步投票器。
+- **Opus 5 仍可使用。** 它保留在 `long-context`，也能通过 `routing.local.yaml` 覆盖任意通道。
+
 ## v0.31.0 新功能
 
 - **目标预算默认不设上限。** `budget_jobs` 和 `budget_seconds` 现在以 JSON `null` 保存并显示为 `unlimited`；此前隐含的 8 个任务和 900 秒上限已移除。只有使用 `--budget-jobs N` 或 `--budget-seconds S` 才会启用硬性上限；重复失败保险丝不是预算，默认仍然启用。
@@ -582,7 +584,7 @@ vendor 一律当成 `work`,而且它只能逐次明确指定,永远不是 lane
   的 Agentic Index 上大幅领先 Flash,而且 2026-07-30 降价后每任务成本只剩零头。
   Flash 只剩吞吐量优势——若你的循环受延迟限制,可在本机覆写把它调回第一。
 - **lane 注释不再放数字。**`routing.yaml` 只说明每条排序「为什么」成立;所有分数、
-  价格与吞吐量连同取数日期,一律放在 `docs/model-capabilities-2026-07.md`。数字过期
+  价格与吞吐量连同取数日期,一律放在 `docs/model-capabilities-2026-09.md`。数字过期
   不再需要动路由表。
 - **新增 value profile**(在 `routing.local.yaml.example`):用约一个 Intelligence
   Index 分数,换每任务成本降三到四成。
@@ -628,7 +630,7 @@ vendor 一律当成 `work`,而且它只能逐次明确指定,永远不是 lane
 - **修复:`routing.local.yaml.example` 还指着已退场的模型**——起手配置里的
   `claude-opus-4-8` 全数改为 `claude-opus-5`(并按通道给对应档位),Gemini 3.5
   Flash 候选改为 3.6 Flash,与 0.10.0 以来的默认值一致。
-- **对照原始资料修正智能指数数字**(`docs/model-capabilities-2026-07.md`):
+- **对照原始资料修正智能指数数字**(`docs/model-capabilities-2026-09.md`):
   那是指数点数不是百分比;补上 AA-Briefcase / GDPval-AA v2 对照,并记下两项与
   默认值相反的结果:Fable 5 在事实知识领先、GPT-5.6 Sol 在呈现质量领先。
 
@@ -665,7 +667,7 @@ vendor 一律当成 `work`,而且它只能逐次明确指定,永远不是 lane
 
 ## v0.9.0 新功能
 
-- **新增 5 个 OpenAI-compatible direct-API vendor** — `deepseek`、`zai`(GLM)、`mistral`、`groq`、`cerebras`,与 `openrouter` 同为免 CLI 通道(curl 加一把 `<VENDOR>_API_KEY`);`lib/common.sh` registry 一行即加一个。详见 [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md)。
+- **新增 5 个 OpenAI-compatible direct-API vendor** — `deepseek`、`zai`(GLM)、`mistral`、`groq`、`cerebras`,与 `openrouter` 同为免 CLI 通道(curl 加一把 `<VENDOR>_API_KEY`);`lib/common.sh` registry 一行即加一个。详见 [`docs/model-capabilities-2026-09.md`](docs/model-capabilities-2026-09.md)。
 - **fish shell 补全** — `omnilane completion fish | source`。
 
 ## v0.8.3 新功能