roadmodel 0.2.98__tar.gz → 0.2.99__tar.gz

{roadmodel-0.2.98 → roadmodel-0.2.99}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: roadmodel
-Version: 0.2.98
+Version: 0.2.99
 Summary: BYO-key CLI that recommends the right AI model, platform, and settings for a prompt.
 Project-URL: Homepage, https://roadmodel.ai
 Project-URL: Repository, https://github.com/nathanramoscfa/roadmodel

{roadmodel-0.2.98/roadmodel/data → roadmodel-0.2.99/docs}/catalog.json

@@ -1,7 +1,7 @@
 {
   "schema_version": "4",
   "output_contract_version": 2,
-  "generated_at_utc": "2026-10-25T19:01:45Z",
+  "generated_at_utc": "2026-10-27T17:02:15Z",
   "source_doc_sha256": {
     "model-selector.txt": "e2ef6c5a18f05ff952fad9f576720b8fc4637f563f60ce5c077c5501fc21d1b7",
     "model-tier-cost-scale.md": "dad707e59d0f3e75cf699d81368ba66783cc2a6a0bf625c430e8f884ef844a1c"

{roadmodel-0.2.98 → roadmodel-0.2.99}/docs/claude-commands/roadmap-refresh.md

@@ -204,8 +204,7 @@ complexity, novel or not; its Model rationale names them), then:
   fix, not a new pick, so it takes no `Settings updated` line.
 - Every ladder, a step's Settings and a phase's Phase roadmap settings
   alike, takes the table layout (roadmodel 0.2.95 on), as `roadmodel
-  ladder` prints it: Cost, Balanced and Quality columns (the one its
-  **Runs:** line names marked ✓), each rank (**Pick**, **Runner‑up**,
+  ladder` prints it: Cost, Balanced and Quality columns, each rank (**Pick**, **Runner‑up**,
   **Backup**) in a column of its own beside its Model / Platform / Level / AA /
   Price rows — Level the bare level of the platform's own dial, AA its AA
   Intelligence Index there (`not measured` where AA has no row), Price its
@@ -222,10 +221,10 @@ complexity, novel or not; its Model rationale names them), then:
   progress` step's is what an agent is running, and a written phase's
   records what wrote its roadmap, so those keep their layout. One
   exception reaches every section: a 0.2.95–0.2.97 table, whose fields are
-  indented with ` ` (some viewers draw each as a highlighted blob),
-  takes the rank column wherever it stands, Complete steps included, every
-  cell kept. That is a layout fix, not a new pick, and takes no `Settings
-  updated` line. Then
+  indented with ` ` (some viewers draw each as a highlighted blob), or
+  a 0.2.98 table still carrying ✓, takes the rank column without ✓ wherever
+  it stands, Complete steps included, every cell kept. That is a layout fix,
+  not a new pick, and takes no `Settings updated` line. Then
   run `roadmodel rerank <every roadmap file>` (roadmodel 0.2.97 on): it
   gives each of those ladder tables the ranks this release computes, so a
   change to how roadmodel ranks reaches every step not yet run. It reads

{roadmodel-0.2.98/roadmodel/data → roadmodel-0.2.99/docs/claude-commands}/roadmap-step.md

@@ -114,8 +114,7 @@ the dials and make sure it is set.
   **Backup**) sits in a column of its own beside its Model, Platform, Level, AA and
   Price rows. Level is the bare level of that platform's own dial (`High`
   is Intelligence High on Codex, Effort High on Claude Code and
-  Antigravity). The ✓ in a column's header marks the one its **Runs:**
-  line names. From roadmodel 0.2.97 the three ranks of a column come from
+  Antigravity). From roadmodel 0.2.97 the three ranks of a column come from
   three makers (a tight pool is a maker's); where too few are funded, a
   missing rank is the Pick one level lower, its Level marked ↓, and the
   note line says which classification the ladder was ranked for. A 0.2.94

{roadmodel-0.2.98/roadmodel/data → roadmodel-0.2.99/docs/templates}/phase-roadmap-template.md

@@ -758,8 +758,7 @@ that owns it.}}
      <c> --complexity <x> [--novel]` prints them for the
      step's classification (MCP tool: `roadmap_settings`).
      Layout (roadmodel 0.2.95 on): one table whose columns
-     are Cost, Balanced and Quality (the column the **Runs:**
-     line names marked ✓). Each rank — **Pick**, **Runner‑up**,
+     are Cost, Balanced and Quality. Each rank — **Pick**, **Runner‑up**,
      **Backup** — names the first of its rows from a column of
      its own (blank below it, no `&nbsp;`), beside
      field rows: Model (bold); Platform; Level, the bare
@@ -774,7 +773,7 @@ that owns it.}}
      no Thinking row: Thinking is always On wherever the
      platform has the toggle, and every AA Index and Price
      already assumes it. A note line under the table explains
-     ✓ and Price, and a `**Conversation:**` line closes the
+     Price, and a `**Conversation:**` line closes the
      ladder. Each rank comes from its own maker (roadmodel
      0.2.97 on), because a tight usage pool is a maker's: the
      Runner-up is the column's recommendation with the Pick's
@@ -892,7 +891,7 @@ step's classification, pasted as printed. The three table
 shapes after it are the fallback where roadmodel cannot be
 reached}}):
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -910,8 +909,8 @@ reached}}):
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -1331,7 +1330,7 @@ printed with its **Runs:** line. Where roadmodel cannot be
 reached, the single-column variant that matches PLATFORM,
 carrying ONLY the rows that platform's panel exposes.}}
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -1349,8 +1348,8 @@ carrying ONLY the rows that platform's panel exposes.}}
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -1481,7 +1480,7 @@ this step's classification, pasted as printed with its
 **Runs:** line; where roadmodel cannot be reached, the
 single-column variant that matches PLATFORM.}}
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -1499,8 +1498,8 @@ single-column variant that matches PLATFORM.}}
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 

{roadmodel-0.2.98/roadmodel/data → roadmodel-0.2.99/docs/templates}/project-roadmap-template.md

@@ -250,7 +250,7 @@ dependency graph in §5 is honoured.
 **Phase roadmap settings** — the session that writes
 `phase01-roadmap.md` (`/roadmap-phase 1`):
 
-|               |          | Cost                | Balanced            | Quality ✓           |
+|               |          | Cost                | Balanced            | Quality             |
 | ------------- | -------- | ------------------- | ------------------- | ------------------- |
 | **Pick**      | Model    | **<Model>**         | **<Model>**         | **<Model>**         |
 |               | Platform | <Platform>          | <Platform>          | <Platform>          |
@@ -268,8 +268,8 @@ dependency graph in §5 is honoured.
 |               | AA       | <AA Index>          | <AA Index>          | <AA Index>          |
 |               | Price    | <$ per 1M>          | <$ per 1M>          | <$ per 1M>          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -327,26 +327,26 @@ key pays for the platform.>
 **Phase roadmap settings** — the session that writes
 `phase02-roadmap.md` (`/roadmap-phase 2`):
 
-|               |          | Cost  | Balanced | Quality ✓ |
-| ------------- | -------- | ----- | -------- | --------- |
-| **Pick**      | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Runner‑up** | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Backup**    | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+|               |          | Cost  | Balanced | Quality |
+| ------------- | -------- | ----- | -------- | ------- |
+| **Pick**      | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Runner‑up** | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Backup**    | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -378,26 +378,26 @@ model uses at that level.
 **Phase roadmap settings** — the session that writes
 `phaseNN-roadmap.md` (`/roadmap-phase N`):
 
-|               |          | Cost  | Balanced | Quality ✓ |
-| ------------- | -------- | ----- | -------- | --------- |
-| **Pick**      | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Runner‑up** | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Backup**    | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+|               |          | Cost  | Balanced | Quality |
+| ------------- | -------- | ----- | -------- | ------- |
+| **Pick**      | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Runner‑up** | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Backup**    | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 

{roadmodel-0.2.98 → roadmodel-0.2.99}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "roadmodel"
-version = "0.2.98"
+version = "0.2.99"
 description = "BYO-key CLI that recommends the right AI model, platform, and settings for a prompt."
 readme = "README.md"
 requires-python = ">=3.11"

{roadmodel-0.2.98/docs → roadmodel-0.2.99/roadmodel/data}/catalog.json

@@ -1,7 +1,7 @@
 {
   "schema_version": "4",
   "output_contract_version": 2,
-  "generated_at_utc": "2026-10-25T19:01:45Z",
+  "generated_at_utc": "2026-10-27T17:02:15Z",
   "source_doc_sha256": {
     "model-selector.txt": "e2ef6c5a18f05ff952fad9f576720b8fc4637f563f60ce5c077c5501fc21d1b7",
     "model-tier-cost-scale.md": "dad707e59d0f3e75cf699d81368ba66783cc2a6a0bf625c430e8f884ef844a1c"

{roadmodel-0.2.98/docs/templates → roadmodel-0.2.99/roadmodel/data}/phase-roadmap-template.md

@@ -758,8 +758,7 @@ that owns it.}}
      <c> --complexity <x> [--novel]` prints them for the
      step's classification (MCP tool: `roadmap_settings`).
      Layout (roadmodel 0.2.95 on): one table whose columns
-     are Cost, Balanced and Quality (the column the **Runs:**
-     line names marked ✓). Each rank — **Pick**, **Runner‑up**,
+     are Cost, Balanced and Quality. Each rank — **Pick**, **Runner‑up**,
      **Backup** — names the first of its rows from a column of
      its own (blank below it, no `&nbsp;`), beside
      field rows: Model (bold); Platform; Level, the bare
@@ -774,7 +773,7 @@ that owns it.}}
      no Thinking row: Thinking is always On wherever the
      platform has the toggle, and every AA Index and Price
      already assumes it. A note line under the table explains
-     ✓ and Price, and a `**Conversation:**` line closes the
+     Price, and a `**Conversation:**` line closes the
      ladder. Each rank comes from its own maker (roadmodel
      0.2.97 on), because a tight usage pool is a maker's: the
      Runner-up is the column's recommendation with the Pick's
@@ -892,7 +891,7 @@ step's classification, pasted as printed. The three table
 shapes after it are the fallback where roadmodel cannot be
 reached}}):
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -910,8 +909,8 @@ reached}}):
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -1331,7 +1330,7 @@ printed with its **Runs:** line. Where roadmodel cannot be
 reached, the single-column variant that matches PLATFORM,
 carrying ONLY the rows that platform's panel exposes.}}
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -1349,8 +1348,8 @@ carrying ONLY the rows that platform's panel exposes.}}
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -1481,7 +1480,7 @@ this step's classification, pasted as printed with its
 **Runs:** line; where roadmodel cannot be reached, the
 single-column variant that matches PLATFORM.}}
 
-|               |          | Cost                  | Balanced ✓            | Quality               |
+|               |          | Cost                  | Balanced              | Quality               |
 | ------------- | -------- | --------------------- | --------------------- | --------------------- |
 | **Pick**      | Model    | **{{Model}}**         | **{{Model}}**         | **{{Model}}**         |
 |               | Platform | {{Platform}}          | {{Platform}}          | {{Platform}}          |
@@ -1499,8 +1498,8 @@ single-column variant that matches PLATFORM.}}
 |               | AA       | {{AA Index}}          | {{AA Index}}          | {{AA Index}}          |
 |               | Price    | {{$ per 1M}}          | {{$ per 1M}}          | {{$ per 1M}}          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 

{roadmodel-0.2.98/docs/templates → roadmodel-0.2.99/roadmodel/data}/project-roadmap-template.md

@@ -250,7 +250,7 @@ dependency graph in §5 is honoured.
 **Phase roadmap settings** — the session that writes
 `phase01-roadmap.md` (`/roadmap-phase 1`):
 
-|               |          | Cost                | Balanced            | Quality ✓           |
+|               |          | Cost                | Balanced            | Quality             |
 | ------------- | -------- | ------------------- | ------------------- | ------------------- |
 | **Pick**      | Model    | **<Model>**         | **<Model>**         | **<Model>**         |
 |               | Platform | <Platform>          | <Platform>          | <Platform>          |
@@ -268,8 +268,8 @@ dependency graph in §5 is honoured.
 |               | AA       | <AA Index>          | <AA Index>          | <AA Index>          |
 |               | Price    | <$ per 1M>          | <$ per 1M>          | <$ per 1M>          |
 
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -327,26 +327,26 @@ key pays for the platform.>
 **Phase roadmap settings** — the session that writes
 `phase02-roadmap.md` (`/roadmap-phase 2`):
 
-|               |          | Cost  | Balanced | Quality ✓ |
-| ------------- | -------- | ----- | -------- | --------- |
-| **Pick**      | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Runner‑up** | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Backup**    | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+|               |          | Cost  | Balanced | Quality |
+| ------------- | -------- | ----- | -------- | ------- |
+| **Pick**      | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Runner‑up** | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Backup**    | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 
@@ -378,26 +378,26 @@ model uses at that level.
 **Phase roadmap settings** — the session that writes
 `phaseNN-roadmap.md` (`/roadmap-phase N`):
 
-|               |          | Cost  | Balanced | Quality ✓ |
-| ------------- | -------- | ----- | -------- | --------- |
-| **Pick**      | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Runner‑up** | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-| **Backup**    | Model    | **…** | **…**    | **…**     |
-|               | Platform | …     | …        | …         |
-|               | Level    | …     | …        | …         |
-|               | AA       | …     | …        | …         |
-|               | Price    | …     | …        | …         |
-
-✓ runs by default. Price: list $ per 1M tokens (3 in : 1 out) × the tokens the
-model uses at that level.
+|               |          | Cost  | Balanced | Quality |
+| ------------- | -------- | ----- | -------- | ------- |
+| **Pick**      | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Runner‑up** | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+| **Backup**    | Model    | **…** | **…**    | **…**   |
+|               | Platform | …     | …        | …       |
+|               | Level    | …     | …        | …       |
+|               | AA       | …     | …        | …       |
+|               | Price    | …     | …        | …       |
+
+Price: list $ per 1M tokens (3 in : 1 out) × the tokens the model uses at that
+level.
 
 **Conversation:** New
 

{roadmodel-0.2.98 → roadmodel-0.2.99}/roadmodel/data/roadmap-refresh.md

@@ -204,8 +204,7 @@ complexity, novel or not; its Model rationale names them), then:
   fix, not a new pick, so it takes no `Settings updated` line.
 - Every ladder, a step's Settings and a phase's Phase roadmap settings
   alike, takes the table layout (roadmodel 0.2.95 on), as `roadmodel
-  ladder` prints it: Cost, Balanced and Quality columns (the one its
-  **Runs:** line names marked ✓), each rank (**Pick**, **Runner‑up**,
+  ladder` prints it: Cost, Balanced and Quality columns, each rank (**Pick**, **Runner‑up**,
   **Backup**) in a column of its own beside its Model / Platform / Level / AA /
   Price rows — Level the bare level of the platform's own dial, AA its AA
   Intelligence Index there (`not measured` where AA has no row), Price its
@@ -222,10 +221,10 @@ complexity, novel or not; its Model rationale names them), then:
   progress` step's is what an agent is running, and a written phase's
   records what wrote its roadmap, so those keep their layout. One
   exception reaches every section: a 0.2.95–0.2.97 table, whose fields are
-  indented with ` ` (some viewers draw each as a highlighted blob),
-  takes the rank column wherever it stands, Complete steps included, every
-  cell kept. That is a layout fix, not a new pick, and takes no `Settings
-  updated` line. Then
+  indented with ` ` (some viewers draw each as a highlighted blob), or
+  a 0.2.98 table still carrying ✓, takes the rank column without ✓ wherever
+  it stands, Complete steps included, every cell kept. That is a layout fix,
+  not a new pick, and takes no `Settings updated` line. Then
   run `roadmodel rerank <every roadmap file>` (roadmodel 0.2.97 on): it
   gives each of those ladder tables the ranks this release computes, so a
   change to how roadmodel ranks reaches every step not yet run. It reads

{roadmodel-0.2.98/docs/claude-commands → roadmodel-0.2.99/roadmodel/data}/roadmap-step.md

@@ -114,8 +114,7 @@ the dials and make sure it is set.
   **Backup**) sits in a column of its own beside its Model, Platform, Level, AA and
   Price rows. Level is the bare level of that platform's own dial (`High`
   is Intelligence High on Codex, Effort High on Claude Code and
-  Antigravity). The ✓ in a column's header marks the one its **Runs:**
-  line names. From roadmodel 0.2.97 the three ranks of a column come from
+  Antigravity). From roadmodel 0.2.97 the three ranks of a column come from
   three makers (a tight pool is a maker's); where too few are funded, a
   missing rank is the Pick one level lower, its Level marked ↓, and the
   note line says which classification the ladder was ranked for. A 0.2.94

{roadmodel-0.2.98 → roadmodel-0.2.99}/src/roadmodel/__init__.py

@@ -1,2 +1,2 @@
 # src/roadmodel/__init__.py
-__version__ = "0.2.98"
+__version__ = "0.2.99"

{roadmodel-0.2.98 → roadmodel-0.2.99}/src/roadmodel/ladder_layout.py

@@ -123,7 +123,7 @@ def _grid(rows: list[list[str]]) -> list[str]:
 def ladder_markdown(
     columns: Mapping[str, Mapping[str, Option | None]],
     *,
-    runs: str | None,
+    runs: str | None = None,
     conversation: str | None,
     ranked_for: str | None = None,
 ) -> str:
@@ -132,7 +132,7 @@ def ladder_markdown(
     none), and a field row only when some option of that rank has a value
     there (an older backup has no AA Index; a rank with no option in any
     column reads `none` alone)."""
-    head = ["", "", *(t.capitalize() + (f" {RUNS_MARK}" if t == runs else "") for t in TIERS)]
+    head = ["", "", *(t.capitalize() for t in TIERS)]
     rows = [head]
     priced = unmeasured = stepped = False
     for label, key in RANKS:
@@ -155,8 +155,6 @@ def ladder_markdown(
         )
         stepped |= any(o is not None and o.stepped_down for o in opts)
     notes = [f"Ranked for {ranked_for}."] if ranked_for else []
-    if runs:
-        notes.append(f"{RUNS_MARK} runs by default.")
     if stepped:
         notes.append(
             f"{STEP_MARK} the Pick one level lower: no other maker is funded, and a "
@@ -357,22 +355,25 @@ def _list_ladder(
 # that runs, its Conversation value, the classification its note names, and
 # whether converting it is a display fix that reaches every section.
 _Found = tuple[int, dict[str, dict[str, "Option | None"]], str | None, str | None, str | None, bool]
-# The 0.2.95–0.2.97 table: one empty header cell before Cost.
-_INDENT_HEAD_RE: Final = re.compile(r"^\|\s*\|\s*Cost\b")
+# The 0.2.95–0.2.98 table: one empty header cell before Cost (0.2.95–0.2.97)
+# or two with a ✓ in the header or note (0.2.98).
+_INDENT_HEAD_RE: Final = re.compile(r"^\|\s*\|(?:\s*\|)?\s*Cost\b")
 _PRICE_RE: Final = re.compile(r"^\$([0-9]+(?:\.[0-9]+)?)/1M(\*?)$")
 
 
 def _indented_ladder(
     lines: list[str], start: int
 ) -> tuple[int, dict[str, dict[str, Option | None]], str | None, str | None, str | None] | None:
-    """A 0.2.95–0.2.97 ladder table starting at ``lines[start]`` (rank rows
-    over ` `-indented fields): the index past its note and Conversation
-    line (its Runs paragraph stays), its columns with every cell read back,
-    the column marked ✓, its Conversation value and the classification its
-    note names; None when a row is not one this release wrote."""
+    """A 0.2.95–0.2.98 ladder table starting at ``lines[start]`` (rank rows
+    over ` `-indented fields, or a 0.2.98 table still carrying ✓): the
+    index past its note and Conversation line (its Runs paragraph stays), its
+    columns with every cell read back, the column marked ✓, its Conversation
+    value and the classification its note names; None when a row is not one
+    those releases wrote or the table is already in the current layout."""
     head = _table_cells(lines[start])
+    tier_cells = head[-len(TIERS) :]
     runs = next(
-        (t for t, cell in zip(TIERS, head[1:], strict=False) if cell.endswith(RUNS_MARK)), None
+        (t for t, cell in zip(TIERS, tier_cells, strict=False) if cell.endswith(RUNS_MARK)), None
     )
     cells, j = _table_rows(lines, start + 2)
     keys = {label.replace("\u2011", "-"): key for label, key in RANKS}
@@ -409,6 +410,8 @@ def _indented_ladder(
             n += 1
         note = " ".join(" ".join(lines[k:n]).split())
         end = n
+    if len(head) == 2 + len(TIERS) and runs is None and f"{RUNS_MARK} runs by default." not in note:
+        return None
     k = _skip_blank(lines, end)
     conversation = None
     if k < len(lines) and (conv := _CONVERSATION_RE.match(lines[k])):

{roadmodel-0.2.98 → roadmodel-0.2.99}/src/roadmodel/scoring.py

@@ -1808,10 +1808,36 @@ def ladder(
     # none: the picks then fall back to the models' frontier.
     points = frontier(settings, bench) or front
 
+    aa_band = _aa_band(bench, cat)
     adequate = [p for p in points if p.candidate.requirement_penalty == 0]
     if not adequate:
         adequate = [max(points, key=_nearest)]
-    top = _top(adequate, task.category)
+    raw_top = _top(adequate, task.category)
+    # Within a maker, a model the evidence cannot tell from the strongest
+    # (within the category and AA noise bands both ways, as in _backup_for)
+    # wins when it is cheaper: GPT-6.1 Sol at $4.00/1M is not passed over for
+    # GPT-6 Astra at $16.36/1M over 1.4 AA points. Each model is read at its
+    # own strongest effort in the band so a model's lower effort does not
+    # displace its higher one on QUALITY.
+    prepaid = raw_top.candidate.scarcity < 1.0
+    peers = [
+        p
+        for p in points
+        if p.candidate.provider == raw_top.candidate.provider
+        and (p.candidate.scarcity < 1.0 or not prepaid)
+        and _near_tied(
+            p.candidate, p.aa_index, raw_top.candidate, raw_top.aa_index, task.category, aa_band
+        )
+    ]
+    by_model: dict[str, list[FrontierPoint]] = {}
+    for p in peers:
+        by_model.setdefault(p.candidate.model_id, []).append(p)
+    top = min(
+        (_top(ps, task.category) for ps in by_model.values()),
+        key=lambda p: (p.price_usd, -p.candidate.score),
+    )
+    if top not in adequate:
+        adequate = [p for p in points if p.candidate.requirement_penalty == 0 or p is top]
     span = adequate[: adequate.index(top) + 1]
     specialist = _specialist(task, settings, points, top, bench)
     by_level = {(c.model_id, c.evidence_level): c for c in settings if c.evidence_level}
@@ -1916,9 +1942,6 @@ def ladder(
         )
     balanced_rung = settle(_below(balanced_rung, quality_rung, cat))
     cost_rung = settle(_below(cost_rung, balanced_rung, cat))
-    # General capability is level within the same noise band, read on the
-    # AA Index's own 0–100 scale (the planning evidence).
-    aa_band = _aa_band(bench, cat)
     rungs = {
         r.tier: replace(r, backup=_backup_for(r, settings, task, headroom, bench, cat, aa_band))
         for r in (cost_rung, balanced_rung, quality_rung)