roadmodel 0.2.78__tar.gz → 0.2.79__tar.gz

{roadmodel-0.2.78 → roadmodel-0.2.79}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: roadmodel
-Version: 0.2.78
+Version: 0.2.79
 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-0.2.78 → roadmodel-0.2.79}/docs/benchmarks.json

@@ -2526,34 +2526,34 @@
       }
     },
     "grok-4.6": {
-      "aa_id": "c8adc5cf-fd5a-407b-af51-dc3bede3e49c",
-      "aa_slug": "grok-4-6",
-      "aa_name": "Grok 4.6 (High)",
+      "aa_id": "d6f43d0a-e8c4-447c-9ff7-5cf0a9d59aa3",
+      "aa_slug": "grok-4-6-xhigh",
+      "aa_name": "Grok 4.6 (Xhigh)",
       "aa_creator": "SpaceXAI",
       "release_date": "2026-08-12",
       "evaluations": {
-        "artificial_analysis_intelligence_index": 44.3,
-        "artificial_analysis_coding_index": 76.8,
+        "artificial_analysis_intelligence_index": 44.2,
+        "artificial_analysis_coding_index": 75.9,
         "artificial_analysis_math_index": null,
         "mmlu_pro": null,
-        "gpqa": 0.949,
-        "hle": 0.429,
+        "gpqa": 0.935,
+        "hle": 0.441,
         "livecodebench": null,
-        "scicode": 0.565,
+        "scicode": 0.53,
         "math_500": null,
         "aime": null,
         "aime_25": null,
         "ifbench": null,
-        "lcr": 0.803333333333333,
+        "lcr": 0.81,
         "terminalbench_hard": null,
-        "terminalbench_v2_1": 0.883895131086142,
-        "terminalbench_v4_0": 0.212121212121212,
+        "terminalbench_v2_1": 0.880149812734082,
+        "terminalbench_v4_0": 0.171717171717172,
         "tau2": null,
-        "tau_banking": 0.507216494845361
+        "tau_banking": 0.43298969072165
       },
       "median_output_tokens_per_second": 0,
       "median_time_to_first_token_seconds": 0,
-      "aa_effort": "high",
+      "aa_effort": "xhigh",
       "effort_variants": {
         "low": {
           "aa_slug": "grok-4-6-low",
@@ -2587,19 +2587,19 @@
           },
           "median_output_tokens_per_second": 0
         },
-        "xhigh": {
-          "aa_slug": "grok-4-6-xhigh",
-          "aa_name": "Grok 4.6 (Xhigh)",
+        "high": {
+          "aa_slug": "grok-4-6",
+          "aa_name": "Grok 4.6 (High)",
           "evaluations": {
-            "artificial_analysis_intelligence_index": 44.2,
-            "artificial_analysis_coding_index": 75.9,
-            "gpqa": 0.935,
-            "hle": 0.441,
-            "scicode": 0.53,
-            "lcr": 0.81,
-            "terminalbench_v2_1": 0.880149812734082,
-            "terminalbench_v4_0": 0.171717171717172,
-            "tau_banking": 0.43298969072165
+            "artificial_analysis_intelligence_index": 44.3,
+            "artificial_analysis_coding_index": 76.8,
+            "gpqa": 0.949,
+            "hle": 0.429,
+            "scicode": 0.565,
+            "lcr": 0.803333333333333,
+            "terminalbench_v2_1": 0.883895131086142,
+            "terminalbench_v4_0": 0.212121212121212,
+            "tau_banking": 0.507216494845361
           },
           "median_output_tokens_per_second": 0
         }

{roadmodel-0.2.78/roadmodel/data → roadmodel-0.2.79/docs}/catalog.json

@@ -1,9 +1,9 @@
 {
   "schema_version": "4",
   "output_contract_version": 2,
-  "generated_at_utc": "2026-10-08T22:10:51Z",
+  "generated_at_utc": "2026-10-09T14:53:46Z",
   "source_doc_sha256": {
-    "model-selector.txt": "58061ed5e6812e7778a4ef31d2337a8c56322202849d68a43d069187133fae2b",
+    "model-selector.txt": "a41aef972265c5acdd4e3826d74f4919ba06c93857f2d0058cb82f7b192c46a4",
     "model-tier-cost-scale.md": "80baddfd88e6a105658eb5dbe64a22d444ff42e55c7b9e42dd3a14c5ec419b89"
   },
   "models": [
@@ -90,7 +90,7 @@
       "cache_read_per_1m": null,
       "tier_cost": "very-high",
       "tiers": {
-        "coding": "A",
+        "coding": "S",
         "planning": "A",
         "agentic": "S",
         "multimodal": "A",
@@ -101,7 +101,7 @@
       "jurisdiction": "us",
       "headline_benchmarks": "AA Intelligence Index 50.8 (max); HLE 54.9% (max); Terminal-Bench 2.1 89.1 (max); \u03c4\u00b2-bench banking pass_1 42.1%",
       "pricing_notes": "Hidden by default; Requires Max Mode on legacy request-based plans; Fast mode (`claude-opus-5-fast`) requires Max Mode on legacy request-based plans; Up to 1M tokens with extended context at the same per-token rates (no long-context surcharge)",
-      "best_for": "Frontier-class agentic work, plus strong coding, planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5.",
+      "best_for": "Frontier-class coding and agentic work, plus strong planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5.",
       "superseded_by": "claude-opus-5-5",
       "superseded_on": "2026-09-24",
       "added_on": null,
@@ -1224,9 +1224,9 @@
         "speed": "B"
       },
       "jurisdiction": "us",
-      "headline_benchmarks": "AA Intelligence Index 44.3 (high); HLE 42.9% (high); Terminal-Bench 2.1 88.4 (high); \u03c4\u00b2-bench banking pass_1 50.7%",
+      "headline_benchmarks": "AA Intelligence Index 44.2 (xhigh); HLE 44.1%; SciCode 53.0; Terminal-Bench 4.0 17.2; AA-LCR 0.810; Terminal-Bench 2.1 88.4 (high); \u03c4\u00b2-bench banking pass_1 43.3%",
       "pricing_notes": "Jointly trained by Cursor and SpaceXAI",
-      "best_for": "Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.3 ranks 13th of 56 measured models. Superseded by Grok 4.7.",
+      "best_for": "Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.2 ranks 13th of 56 measured models. Superseded by Grok 4.7.",
       "superseded_by": "grok-4.7",
       "superseded_on": "2026-09-24",
       "added_on": null,

{roadmodel-0.2.78 → roadmodel-0.2.79}/docs/effort-tokens.json

@@ -1,6 +1,6 @@
 {
   "schema_version": 1,
-  "generated_at_utc": "2026-10-09T05:01:40Z",
+  "generated_at_utc": "2026-10-09T14:32:00Z",
   "source": "Measured by update/measure_effort_tokens.py: output tokens (reasoning included) per probe on a fixed set of short tasks, at each effort Artificial Analysis measured the model at.",
   "probe_set_sha256": "db1a808e774b2c1afd25f380c9641b83ccacae87f1ead88fc7307f2c539aa4a0",
   "probes": 8,
@@ -779,10 +779,26 @@
           "multiplier": 1.207
         }
       }
+    },
+    "muse-spark-1.3": {
+      "lane": "openrouter",
+      "levels": {
+        "xhigh": {
+          "mean_output_tokens": 2151.3,
+          "runs": 16,
+          "truncated": 0,
+          "multiplier": 2.449
+        },
+        "max": {
+          "mean_output_tokens": 2058.7,
+          "runs": 16,
+          "truncated": 0,
+          "multiplier": 2.449
+        }
+      }
     }
   },
   "unmeasured": {
-    "gemini-3-pro": "no lane serves it (not on OpenRouter, no subscription lane)",
-    "muse-spark-1.3": "runs failed: {'message': 'This model requires you to complete the following before use: 18+ age confirmation. Confirm at https://openrouter.ai/settings/preferences.', 'code'"
+    "gemini-3-pro": "no lane serves it (not on OpenRouter, no subscription lane)"
   }
 }

{roadmodel-0.2.78 → roadmodel-0.2.79}/docs/model-selector.txt

@@ -982,13 +982,13 @@
       <model id="claude-opus-5" name="Opus 5"
              input-price-per-1m="$5.00" output-price-per-1m="$25.00"
              jurisdiction="us"
-             tier-coding="A" tier-planning="A" tier-agentic="S"
+             tier-coding="S" tier-planning="A" tier-agentic="S"
              tier-multimodal="A" tier-long-context="A" tier-knowledge="A"
              tier-speed="D"
              headline-benchmarks="AA Intelligence Index 50.8 (max); HLE 54.9% (max); Terminal-Bench 2.1 89.1 (max); τ²-bench banking pass_1 42.1%"
              pricing-notes="Hidden by default; Requires Max Mode on legacy request-based plans; Fast mode (`claude-opus-5-fast`) requires Max Mode on legacy request-based plans; Up to 1M tokens with extended context at the same per-token rates (no long-context surcharge)"
              prose-generated="best-for"
-             best-for="Frontier-class agentic work, plus strong coding, planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5." superseded-by="claude-opus-5-5" superseded-on="2026-09-24"/>
+             best-for="Frontier-class coding and agentic work, plus strong planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5." superseded-by="claude-opus-5-5" superseded-on="2026-09-24"/>
       <model id="claude-fable-5" name="Fable 5"
              input-price-per-1m="$10.00" output-price-per-1m="$50.00"
              jurisdiction="us"
@@ -1541,10 +1541,10 @@
              tier-coding="A" tier-planning="A" tier-agentic="A"
              tier-multimodal="B" tier-long-context="A" tier-knowledge="A"
              tier-speed="B"
-             headline-benchmarks="AA Intelligence Index 44.3 (high); HLE 42.9% (high); Terminal-Bench 2.1 88.4 (high); τ²-bench banking pass_1 50.7%"
+             headline-benchmarks="AA Intelligence Index 44.2 (xhigh); HLE 44.1%; SciCode 53.0; Terminal-Bench 4.0 17.2; AA-LCR 0.810; Terminal-Bench 2.1 88.4 (high); τ²-bench banking pass_1 43.3%"
              pricing-notes="Jointly trained by Cursor and SpaceXAI"
-             prose-generated="best-for"
-             best-for="Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.3 ranks 13th of 56 measured models. Superseded by Grok 4.7." superseded-by="grok-4.7" superseded-on="2026-09-24"/>
+             prose-generated="best-for headline-benchmarks"
+             best-for="Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.2 ranks 13th of 56 measured models. Superseded by Grok 4.7." superseded-by="grok-4.7" superseded-on="2026-09-24"/>
     </tier>
   </model-options>
 
@@ -1870,7 +1870,7 @@
         context (opus-4.7 1M, opus-4.8 1M, claude-opus-5 1M, claude-opus-5-5 1M, gemini-3.1-pro 1M, gemini-3.5-flash-lite 1M) over forcing
         a smaller-context model into Max Mode truncation.
       - For PRIMARY = `coding` at S-tier requirement, the candidate set is
-        muse-spark-1.3, glm-5.3, claude-sonnet-5-5, gpt-6-sol, claude-opus-5-5, claude-fable-5, claude-fable-5.1, gpt-6-astra; cost tie-breaker favors
+        muse-spark-1.3, glm-5.3, claude-sonnet-5-5, gpt-6-sol, claude-opus-5-5, claude-opus-5, claude-fable-5, claude-fable-5.1, gpt-6-astra; cost tie-breaker favors
         muse-spark-1.3 when the ratings are equivalent for the prompt.
       - Default to composer-2.5 for routine multi-file implementation when a
         coding-A rating suffices; escalate only on a concrete capability

{roadmodel-0.2.78 → roadmodel-0.2.79}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "roadmodel"
-version = "0.2.78"
+version = "0.2.79"
 description = "BYO-key CLI that recommends the right AI model, platform, and settings for a prompt."
 readme = "README.md"
 requires-python = ">=3.11"

{roadmodel-0.2.78 → roadmodel-0.2.79}/roadmodel/data/benchmarks.json

@@ -2526,34 +2526,34 @@
       }
     },
     "grok-4.6": {
-      "aa_id": "c8adc5cf-fd5a-407b-af51-dc3bede3e49c",
-      "aa_slug": "grok-4-6",
-      "aa_name": "Grok 4.6 (High)",
+      "aa_id": "d6f43d0a-e8c4-447c-9ff7-5cf0a9d59aa3",
+      "aa_slug": "grok-4-6-xhigh",
+      "aa_name": "Grok 4.6 (Xhigh)",
       "aa_creator": "SpaceXAI",
       "release_date": "2026-08-12",
       "evaluations": {
-        "artificial_analysis_intelligence_index": 44.3,
-        "artificial_analysis_coding_index": 76.8,
+        "artificial_analysis_intelligence_index": 44.2,
+        "artificial_analysis_coding_index": 75.9,
         "artificial_analysis_math_index": null,
         "mmlu_pro": null,
-        "gpqa": 0.949,
-        "hle": 0.429,
+        "gpqa": 0.935,
+        "hle": 0.441,
         "livecodebench": null,
-        "scicode": 0.565,
+        "scicode": 0.53,
         "math_500": null,
         "aime": null,
         "aime_25": null,
         "ifbench": null,
-        "lcr": 0.803333333333333,
+        "lcr": 0.81,
         "terminalbench_hard": null,
-        "terminalbench_v2_1": 0.883895131086142,
-        "terminalbench_v4_0": 0.212121212121212,
+        "terminalbench_v2_1": 0.880149812734082,
+        "terminalbench_v4_0": 0.171717171717172,
         "tau2": null,
-        "tau_banking": 0.507216494845361
+        "tau_banking": 0.43298969072165
       },
       "median_output_tokens_per_second": 0,
       "median_time_to_first_token_seconds": 0,
-      "aa_effort": "high",
+      "aa_effort": "xhigh",
       "effort_variants": {
         "low": {
           "aa_slug": "grok-4-6-low",
@@ -2587,19 +2587,19 @@
           },
           "median_output_tokens_per_second": 0
         },
-        "xhigh": {
-          "aa_slug": "grok-4-6-xhigh",
-          "aa_name": "Grok 4.6 (Xhigh)",
+        "high": {
+          "aa_slug": "grok-4-6",
+          "aa_name": "Grok 4.6 (High)",
           "evaluations": {
-            "artificial_analysis_intelligence_index": 44.2,
-            "artificial_analysis_coding_index": 75.9,
-            "gpqa": 0.935,
-            "hle": 0.441,
-            "scicode": 0.53,
-            "lcr": 0.81,
-            "terminalbench_v2_1": 0.880149812734082,
-            "terminalbench_v4_0": 0.171717171717172,
-            "tau_banking": 0.43298969072165
+            "artificial_analysis_intelligence_index": 44.3,
+            "artificial_analysis_coding_index": 76.8,
+            "gpqa": 0.949,
+            "hle": 0.429,
+            "scicode": 0.565,
+            "lcr": 0.803333333333333,
+            "terminalbench_v2_1": 0.883895131086142,
+            "terminalbench_v4_0": 0.212121212121212,
+            "tau_banking": 0.507216494845361
           },
           "median_output_tokens_per_second": 0
         }

{roadmodel-0.2.78/docs → roadmodel-0.2.79/roadmodel/data}/catalog.json

@@ -1,9 +1,9 @@
 {
   "schema_version": "4",
   "output_contract_version": 2,
-  "generated_at_utc": "2026-10-08T22:10:51Z",
+  "generated_at_utc": "2026-10-09T14:53:46Z",
   "source_doc_sha256": {
-    "model-selector.txt": "58061ed5e6812e7778a4ef31d2337a8c56322202849d68a43d069187133fae2b",
+    "model-selector.txt": "a41aef972265c5acdd4e3826d74f4919ba06c93857f2d0058cb82f7b192c46a4",
     "model-tier-cost-scale.md": "80baddfd88e6a105658eb5dbe64a22d444ff42e55c7b9e42dd3a14c5ec419b89"
   },
   "models": [
@@ -90,7 +90,7 @@
       "cache_read_per_1m": null,
       "tier_cost": "very-high",
       "tiers": {
-        "coding": "A",
+        "coding": "S",
         "planning": "A",
         "agentic": "S",
         "multimodal": "A",
@@ -101,7 +101,7 @@
       "jurisdiction": "us",
       "headline_benchmarks": "AA Intelligence Index 50.8 (max); HLE 54.9% (max); Terminal-Bench 2.1 89.1 (max); \u03c4\u00b2-bench banking pass_1 42.1%",
       "pricing_notes": "Hidden by default; Requires Max Mode on legacy request-based plans; Fast mode (`claude-opus-5-fast`) requires Max Mode on legacy request-based plans; Up to 1M tokens with extended context at the same per-token rates (no long-context surcharge)",
-      "best_for": "Frontier-class agentic work, plus strong coding, planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5.",
+      "best_for": "Frontier-class coding and agentic work, plus strong planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5.",
       "superseded_by": "claude-opus-5-5",
       "superseded_on": "2026-09-24",
       "added_on": null,
@@ -1224,9 +1224,9 @@
         "speed": "B"
       },
       "jurisdiction": "us",
-      "headline_benchmarks": "AA Intelligence Index 44.3 (high); HLE 42.9% (high); Terminal-Bench 2.1 88.4 (high); \u03c4\u00b2-bench banking pass_1 50.7%",
+      "headline_benchmarks": "AA Intelligence Index 44.2 (xhigh); HLE 44.1%; SciCode 53.0; Terminal-Bench 4.0 17.2; AA-LCR 0.810; Terminal-Bench 2.1 88.4 (high); \u03c4\u00b2-bench banking pass_1 43.3%",
       "pricing_notes": "Jointly trained by Cursor and SpaceXAI",
-      "best_for": "Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.3 ranks 13th of 56 measured models. Superseded by Grok 4.7.",
+      "best_for": "Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.2 ranks 13th of 56 measured models. Superseded by Grok 4.7.",
       "superseded_by": "grok-4.7",
       "superseded_on": "2026-09-24",
       "added_on": null,

{roadmodel-0.2.78 → roadmodel-0.2.79}/roadmodel/data/effort-tokens.json

@@ -1,6 +1,6 @@
 {
   "schema_version": 1,
-  "generated_at_utc": "2026-10-09T05:01:40Z",
+  "generated_at_utc": "2026-10-09T14:32:00Z",
   "source": "Measured by update/measure_effort_tokens.py: output tokens (reasoning included) per probe on a fixed set of short tasks, at each effort Artificial Analysis measured the model at.",
   "probe_set_sha256": "db1a808e774b2c1afd25f380c9641b83ccacae87f1ead88fc7307f2c539aa4a0",
   "probes": 8,
@@ -779,10 +779,26 @@
           "multiplier": 1.207
         }
       }
+    },
+    "muse-spark-1.3": {
+      "lane": "openrouter",
+      "levels": {
+        "xhigh": {
+          "mean_output_tokens": 2151.3,
+          "runs": 16,
+          "truncated": 0,
+          "multiplier": 2.449
+        },
+        "max": {
+          "mean_output_tokens": 2058.7,
+          "runs": 16,
+          "truncated": 0,
+          "multiplier": 2.449
+        }
+      }
     }
   },
   "unmeasured": {
-    "gemini-3-pro": "no lane serves it (not on OpenRouter, no subscription lane)",
-    "muse-spark-1.3": "runs failed: {'message': 'This model requires you to complete the following before use: 18+ age confirmation. Confirm at https://openrouter.ai/settings/preferences.', 'code'"
+    "gemini-3-pro": "no lane serves it (not on OpenRouter, no subscription lane)"
   }
 }

{roadmodel-0.2.78 → roadmodel-0.2.79}/roadmodel/data/model-selector.txt

@@ -982,13 +982,13 @@
       <model id="claude-opus-5" name="Opus 5"
              input-price-per-1m="$5.00" output-price-per-1m="$25.00"
              jurisdiction="us"
-             tier-coding="A" tier-planning="A" tier-agentic="S"
+             tier-coding="S" tier-planning="A" tier-agentic="S"
              tier-multimodal="A" tier-long-context="A" tier-knowledge="A"
              tier-speed="D"
              headline-benchmarks="AA Intelligence Index 50.8 (max); HLE 54.9% (max); Terminal-Bench 2.1 89.1 (max); τ²-bench banking pass_1 42.1%"
              pricing-notes="Hidden by default; Requires Max Mode on legacy request-based plans; Fast mode (`claude-opus-5-fast`) requires Max Mode on legacy request-based plans; Up to 1M tokens with extended context at the same per-token rates (no long-context surcharge)"
              prose-generated="best-for"
-             best-for="Frontier-class agentic work, plus strong coding, planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5." superseded-by="claude-opus-5-5" superseded-on="2026-09-24"/>
+             best-for="Frontier-class coding and agentic work, plus strong planning, multimodal input, long context and knowledge, at a very high price. Its AA Intelligence Index of 50.8 ranks 6th of 56 measured models. Weakest at speed (D). Superseded by Opus 5.5." superseded-by="claude-opus-5-5" superseded-on="2026-09-24"/>
       <model id="claude-fable-5" name="Fable 5"
              input-price-per-1m="$10.00" output-price-per-1m="$50.00"
              jurisdiction="us"
@@ -1541,10 +1541,10 @@
              tier-coding="A" tier-planning="A" tier-agentic="A"
              tier-multimodal="B" tier-long-context="A" tier-knowledge="A"
              tier-speed="B"
-             headline-benchmarks="AA Intelligence Index 44.3 (high); HLE 42.9% (high); Terminal-Bench 2.1 88.4 (high); τ²-bench banking pass_1 50.7%"
+             headline-benchmarks="AA Intelligence Index 44.2 (xhigh); HLE 44.1%; SciCode 53.0; Terminal-Bench 4.0 17.2; AA-LCR 0.810; Terminal-Bench 2.1 88.4 (high); τ²-bench banking pass_1 43.3%"
              pricing-notes="Jointly trained by Cursor and SpaceXAI"
-             prose-generated="best-for"
-             best-for="Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.3 ranks 13th of 56 measured models. Superseded by Grok 4.7." superseded-by="grok-4.7" superseded-on="2026-09-24"/>
+             prose-generated="best-for headline-benchmarks"
+             best-for="Strong coding, planning, agentic work, long context and knowledge at a low price. Its AA Intelligence Index of 44.2 ranks 13th of 56 measured models. Superseded by Grok 4.7." superseded-by="grok-4.7" superseded-on="2026-09-24"/>
     </tier>
   </model-options>
 
@@ -1870,7 +1870,7 @@
         context (opus-4.7 1M, opus-4.8 1M, claude-opus-5 1M, claude-opus-5-5 1M, gemini-3.1-pro 1M, gemini-3.5-flash-lite 1M) over forcing
         a smaller-context model into Max Mode truncation.
       - For PRIMARY = `coding` at S-tier requirement, the candidate set is
-        muse-spark-1.3, glm-5.3, claude-sonnet-5-5, gpt-6-sol, claude-opus-5-5, claude-fable-5, claude-fable-5.1, gpt-6-astra; cost tie-breaker favors
+        muse-spark-1.3, glm-5.3, claude-sonnet-5-5, gpt-6-sol, claude-opus-5-5, claude-opus-5, claude-fable-5, claude-fable-5.1, gpt-6-astra; cost tie-breaker favors
         muse-spark-1.3 when the ratings are equivalent for the prompt.
       - Default to composer-2.5 for routine multi-file implementation when a
         coding-A rating suffices; escalate only on a concrete capability

{roadmodel-0.2.78 → roadmodel-0.2.79}/src/roadmodel/__init__.py

@@ -1,2 +1,2 @@
 # src/roadmodel/__init__.py
-__version__ = "0.2.78"
+__version__ = "0.2.79"

{roadmodel-0.2.78 → roadmodel-0.2.79}/src/roadmodel/scoring.py

@@ -101,7 +101,7 @@ LEVEL_ORDER: Final[tuple[str, ...]] = (
 # web catalog's too (tests/test_derived_ratings.py holds the three together).
 # ``speed`` reads the top-level tokens/s field: it enters quality here, but
 # its letter stays an estimate, since AA measures throughput on the maker's
-# own endpoint at max effort. ``None`` = the letter only (multimodal: AA runs
+# own endpoint at the headline (top measured) effort. ``None`` = the letter only (multimodal: AA runs
 # no multimodal evaluation). A ``COMPOSITES`` key is computed by
 # ``with_composites``.
 CATEGORY_EVIDENCE: Final[dict[str, str | None]] = {