multi-forge 1.0.0__tar.gz → 1.0.1__tar.gz

{multi_forge-1.0.0 → multi_forge-1.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: multi-forge
-Version: 1.0.0
+Version: 1.0.1
 Summary: Multi-runtime agent toolkit: proxy routing, cost control, session management, policy enforcement, and workflow orchestration
 Project-URL: Homepage, https://github.com/hapa1i/multi-forge
 Project-URL: Repository, https://github.com/hapa1i/multi-forge
@@ -33,7 +33,7 @@ Requires-Dist: expression<6,>=5.6.0
 Requires-Dist: fastapi-sso<1,>=0.19.0
 Requires-Dist: fastapi<0.137,>=0.136.3
 Requires-Dist: httpx>=0.25.0
-Requires-Dist: litellm<=1.99.0,>=1.98.0
+Requires-Dist: litellm==1.102.0
 Requires-Dist: openai>=1.0.0
 Requires-Dist: orjson<4,>=3.11.6
 Requires-Dist: packaging>=21.0

{multi_forge-1.0.0 → multi_forge-1.0.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "multi-forge"
-version = "1.0.0"
+version = "1.0.1"
 description = "Multi-runtime agent toolkit: proxy routing, cost control, session management, policy enforcement, and workflow orchestration"
 readme = "README.md"
 license = "Apache-2.0"
@@ -42,7 +42,7 @@ dependencies = [
     # Proxy dependencies (previously optional, now core). Forge owns the
     # default-Uvicorn LiteLLM startup set because the upstream `proxy` extra
     # caps cryptography below Forge's patched floor. This set is start/health
-    # validated through LiteLLM 1.99.0; run scripts/test-wheel-runtime.sh before
+    # validated through LiteLLM 1.102.0; run scripts/test-wheel-runtime.sh before
     # raising the ceiling. Alternate servers, databases, and cloud integrations
     # remain outside Forge's runtime path.
     "aiohttp>=3.14.3,<4",
@@ -68,10 +68,8 @@ dependencies = [
     "uvloop>=0.21.0,<1; sys_platform != 'win32'",
     "watchfiles>=0.20",
     "websockets>=15.0.1,<16",
-    # Floor = the first packaged-cost-map version for gemini-3.7-flash (see
-    # tests/integration/proxy/test_proxy_local_litellm_e2e.py); packaged
-    # cost-map pricing for that model first ships in the v1.98 line.
-    "litellm>=1.98.0,<=1.99.0",
+    # Packaged Gemini 3.8 Flash metadata and the verified proxy runtime.
+    "litellm==1.102.0",
 ]
 
 [project.scripts]

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/backends/litellm.yaml

@@ -6,14 +6,15 @@
 #     - gemini/gemini-3.1-pro-preview
 #     - gemini/gemini-3.1-pro-preview-customtools (optimized for custom tools + bash)
 #     - gemini/gemini-3-flash-preview
-#     - gemini/gemini-3.7-flash (LiteLLM Flash default; 3.8 is OpenRouter-only for now)
+#     - gemini/gemini-3.8-flash (LiteLLM Flash default)
+#     - gemini/gemini-3.7-flash
 #     - gemini/gemini-3.6-flash (Gemini 3.6 Flash - efficiency release)
 #     - gemini/gemini-3.5-flash
 #     - gemini/gemini-2.5-flash
 #     - gemini/gemini-2.5-pro
 #
 #   Anthropic (requires ANTHROPIC_API_KEY):
-#     - anthropic/claude-haiku-4-5, claude-sonnet-5, claude-opus-5
+#     - anthropic/claude-haiku-4-5, claude-sonnet-5, claude-opus-5-5, claude-opus-5
 #     - anthropic/claude-fable-5-1, claude-fable-5
 #     - anthropic/claude-opus-4-8, claude-opus-4-6, claude-sonnet-4-6
 #
@@ -25,7 +26,7 @@
 #     - openai/gpt-5.4, gpt-5.4-pro
 #     - openai/gpt-5.5
 #     - openai/gpt-5.6, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna
-#     - openai/gpt-6-astra
+#     - openai/gpt-6-astra, gpt-6-sol, gpt-6-luna
 #
 # Start with: forge model backend start litellm --port 4000
 
@@ -48,7 +49,13 @@ model_list:
       model: gemini/gemini-3-flash-preview
       api_key: os.environ/GEMINI_API_KEY
 
-  # Gemini 3.7 Flash - current LiteLLM Flash default
+  # Gemini 3.8 Flash - current LiteLLM Flash default
+  - model_name: gemini/gemini-3.8-flash
+    litellm_params:
+      model: gemini/gemini-3.8-flash
+      api_key: os.environ/GEMINI_API_KEY
+
+  # Gemini 3.7 Flash
   - model_name: gemini/gemini-3.7-flash
     litellm_params:
       model: gemini/gemini-3.7-flash
@@ -94,6 +101,11 @@ model_list:
       model: gemini/gemini-3-flash-preview
       api_key: os.environ/GEMINI_API_KEY
 
+  - model_name: gemini-3.8-flash
+    litellm_params:
+      model: gemini/gemini-3.8-flash
+      api_key: os.environ/GEMINI_API_KEY
+
   - model_name: gemini-3.7-flash
     litellm_params:
       model: gemini/gemini-3.7-flash
@@ -133,6 +145,45 @@ model_list:
       model: anthropic/claude-sonnet-5
       api_key: os.environ/ANTHROPIC_API_KEY
 
+  # LiteLLM 1.102 predates Opus 5.5. Keep its pricing and thinking contract available offline.
+  # https://platform.claude.com/docs/en/models/opus-5-5/overview (verified 2026-09-23)
+  - model_name: anthropic/claude-opus-5-5
+    litellm_params:
+      model: anthropic/claude-opus-5-5
+      api_key: os.environ/ANTHROPIC_API_KEY
+    model_info:
+      base_model: claude-opus-5-5
+      mode: chat
+      max_input_tokens: 1000000
+      max_output_tokens: 128000
+      max_tokens: 128000
+      input_cost_per_token: 0.000004
+      output_cost_per_token: 0.000020
+      cache_creation_input_token_cost: 0.000005
+      cache_creation_input_token_cost_above_1hr: 0.000008
+      # Opus 5.5 cache reads intentionally cost $0.20/MTok (0.05x base input).
+      cache_read_input_token_cost: 0.0000002
+      supports_vision: true
+      supports_function_calling: true
+      supports_tool_choice: true
+      supports_prompt_caching: true
+      prompt_cache_min_tokens: 512
+      supports_reasoning: true
+      supports_adaptive_thinking: true
+      supports_legacy_thinking: false
+      thinking_always_on: true
+      supports_output_config: true
+      supports_xhigh_reasoning_effort: true
+      supports_max_reasoning_effort: true
+      default_reasoning_effort: medium
+      reasoning_effort_levels: [low, medium, high, xhigh, max]
+      supports_response_schema: true
+      supports_native_structured_output: true
+      supports_assistant_prefill: false
+      provider_specific_entry:
+        us: 1.1
+        fast: 2.0
+
   - model_name: anthropic/claude-opus-5
     litellm_params:
       model: anthropic/claude-opus-5
@@ -284,24 +335,111 @@ model_list:
       model: openai/gpt-5.6-luna
       api_key: os.environ/OPENAI_API_KEY
 
-  # LiteLLM 1.99 lacks packaged Astra pricing. Keep this route usable without a cost-map refresh.
-  # Remove the pricing override once LiteLLM bundles it; test_astra_pricing pins this boundary.
-  # Standard rates, including the >272k-token premium, verified 2026-09-05:
-  # https://developers.openai.com/api/docs/models/gpt-6-astra
   - model_name: openai/gpt-6-astra
     litellm_params:
       model: openai/gpt-6-astra
       api_key: os.environ/OPENAI_API_KEY
+
+  # LiteLLM 1.102 predates Sol and Luna. Standard, Flex and Fast rates include the >272k-token premium.
+  # Flex costs half of standard; Fast (also called priority) costs twice the applicable rate.
+  # Remove these overrides when bundled metadata covers them; test_litellm_model_metadata pins that boundary.
+  # https://developers.openai.com/api/docs/models/gpt-6-sol (verified 2026-09-23)
+  - model_name: openai/gpt-6-sol
+    litellm_params:
+      model: openai/gpt-6-sol
+      api_key: os.environ/OPENAI_API_KEY
+      use_responses_api: true
+    model_info:
+      base_model: gpt-6-sol
+      mode: responses
+      max_input_tokens: 922000
+      max_output_tokens: 128000
+      max_tokens: 128000
+      input_cost_per_token: 0.000002
+      output_cost_per_token: 0.000010
+      cache_read_input_token_cost: 0.0000002
+      cache_creation_input_token_cost: 0.0000025
+      input_cost_per_token_above_272k_tokens: 0.000004
+      output_cost_per_token_above_272k_tokens: 0.000015
+      cache_read_input_token_cost_above_272k_tokens: 0.0000004
+      cache_creation_input_token_cost_above_272k_tokens: 0.000005
+      input_cost_per_token_flex: 0.000001
+      output_cost_per_token_flex: 0.000005
+      cache_read_input_token_cost_flex: 0.0000001
+      cache_creation_input_token_cost_flex: 0.00000125
+      input_cost_per_token_above_272k_tokens_flex: 0.000002
+      output_cost_per_token_above_272k_tokens_flex: 0.0000075
+      cache_read_input_token_cost_above_272k_tokens_flex: 0.0000002
+      cache_creation_input_token_cost_above_272k_tokens_flex: 0.0000025
+      input_cost_per_token_priority: 0.000004
+      output_cost_per_token_priority: 0.000020
+      cache_read_input_token_cost_priority: 0.0000004
+      cache_creation_input_token_cost_priority: 0.000005
+      input_cost_per_token_above_272k_tokens_priority: 0.000008
+      output_cost_per_token_above_272k_tokens_priority: 0.000030
+      cache_read_input_token_cost_above_272k_tokens_priority: 0.0000008
+      cache_creation_input_token_cost_above_272k_tokens_priority: 0.000010
+      supports_vision: true
+      supports_function_calling: true
+      supports_parallel_function_calling: true
+      supports_tool_choice: true
+      supports_prompt_caching: true
+      supports_reasoning: true
+      supports_none_reasoning_effort: true
+      supports_xhigh_reasoning_effort: true
+      supports_max_reasoning_effort: true
+      default_reasoning_effort: medium
+      reasoning_effort_levels: [none, low, medium, high, xhigh, max]
+      supports_response_schema: true
+
+  # https://developers.openai.com/api/docs/models/gpt-6-luna (verified 2026-09-23)
+  - model_name: openai/gpt-6-luna
+    litellm_params:
+      model: openai/gpt-6-luna
+      api_key: os.environ/OPENAI_API_KEY
+      use_responses_api: true
     model_info:
-      base_model: gpt-6-astra
-      input_cost_per_token: 0.000010
-      output_cost_per_token: 0.000050
-      cache_read_input_token_cost: 0.000001
-      cache_creation_input_token_cost: 0.0000125
-      input_cost_per_token_above_272k_tokens: 0.000020
-      output_cost_per_token_above_272k_tokens: 0.000075
-      cache_read_input_token_cost_above_272k_tokens: 0.000002
-      cache_creation_input_token_cost_above_272k_tokens: 0.000025
+      base_model: gpt-6-luna
+      mode: responses
+      max_input_tokens: 922000
+      max_output_tokens: 128000
+      max_tokens: 128000
+      input_cost_per_token: 0.0000001
+      output_cost_per_token: 0.0000005
+      cache_read_input_token_cost: 0.00000001
+      cache_creation_input_token_cost: 0.000000125
+      input_cost_per_token_above_272k_tokens: 0.0000002
+      output_cost_per_token_above_272k_tokens: 0.00000075
+      cache_read_input_token_cost_above_272k_tokens: 0.00000002
+      cache_creation_input_token_cost_above_272k_tokens: 0.00000025
+      input_cost_per_token_flex: 0.00000005
+      output_cost_per_token_flex: 0.00000025
+      cache_read_input_token_cost_flex: 0.000000005
+      cache_creation_input_token_cost_flex: 0.0000000625
+      input_cost_per_token_above_272k_tokens_flex: 0.0000001
+      output_cost_per_token_above_272k_tokens_flex: 0.000000375
+      cache_read_input_token_cost_above_272k_tokens_flex: 0.00000001
+      cache_creation_input_token_cost_above_272k_tokens_flex: 0.000000125
+      input_cost_per_token_priority: 0.0000002
+      output_cost_per_token_priority: 0.000001
+      cache_read_input_token_cost_priority: 0.00000002
+      cache_creation_input_token_cost_priority: 0.00000025
+      input_cost_per_token_above_272k_tokens_priority: 0.0000004
+      output_cost_per_token_above_272k_tokens_priority: 0.0000015
+      cache_read_input_token_cost_above_272k_tokens_priority: 0.00000004
+      cache_creation_input_token_cost_above_272k_tokens_priority: 0.0000005
+      supports_vision: true
+      supports_function_calling: true
+      supports_parallel_function_calling: true
+      supports_tool_choice: true
+      supports_prompt_caching: true
+      supports_reasoning: true
+      supports_none_reasoning_effort: true
+      supports_xhigh_reasoning_effort: true
+      supports_max_reasoning_effort: true
+      default_reasoning_effort: medium
+      reasoning_effort_levels: [none, low, medium, high, xhigh, max]
+      supports_response_schema: true
 
 litellm_settings:
   # Drop unmapped params instead of erroring

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/anthropic-passthrough.yaml

@@ -28,4 +28,4 @@ proxy:
     tiers:
       haiku: claude-haiku-4-5
       sonnet: claude-sonnet-5
-      opus: claude-opus-5
+      opus: claude-opus-5-5

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/codex-responses-local.yaml

@@ -27,7 +27,13 @@ proxy:
       sonnet: openai/gpt-6-astra
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
       sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-anthropic-local.yaml

@@ -6,8 +6,8 @@
 #
 # Backend (LiteLLM on port 4000) starts automatically.
 #
-# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5. Alternatives are
-# claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
+# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5.5. Alternatives are
+# claude-opus-5, claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
 # and claude-sonnet-4-6.
 #
 # After adding ANTHROPIC_API_KEY, stop the backend so Forge restarts it with the new environment:
@@ -23,9 +23,10 @@ proxy:
     tiers:
       haiku: anthropic/claude-haiku-4-5
       sonnet: anthropic/claude-sonnet-5
-      opus: anthropic/claude-opus-5
+      opus: anthropic/claude-opus-5-5
     model_alternatives:
       opus:
+        claude-opus-5: anthropic/claude-opus-5
         claude-fable-5-1: anthropic/claude-fable-5-1
         claude-fable-5: anthropic/claude-fable-5
         claude-opus-4-8: anthropic/claude-opus-4-8

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-anthropic.yaml

@@ -1,12 +1,12 @@
 # Template: litellm-anthropic
 # Anthropic Claude via remote/shared LiteLLM
 #
-# Access to Claude Opus 5, Sonnet 5, Fable 5.1/5, Opus 4.8/4.6, Sonnet 4.6, and Haiku 4.5.
+# Access to Claude Opus 5.5/5, Sonnet 5, Fable 5.1/5, Opus 4.8/4.6, Sonnet 4.6, and Haiku 4.5.
 # Requires LITELLM_API_KEY and LITELLM_BASE_URL in .env (or environment).
-# NOTE: the shared LiteLLM instance must serve anthropic/claude-opus-5 and anthropic/claude-sonnet-5.
+# NOTE: the shared LiteLLM instance must serve anthropic/claude-opus-5-5 and anthropic/claude-sonnet-5.
 #
-# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5. Alternatives are
-# claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
+# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5.5. Alternatives are
+# claude-opus-5, claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
 # and claude-sonnet-4-6.
 # prompt_caching: passthrough respects Anthropic's 4-breakpoint cache_control limit.
 
@@ -20,9 +20,10 @@ proxy:
     tiers:
       haiku: anthropic/claude-haiku-4-5
       sonnet: anthropic/claude-sonnet-5
-      opus: anthropic/claude-opus-5
+      opus: anthropic/claude-opus-5-5
     model_alternatives:
       opus:
+        claude-opus-5: anthropic/claude-opus-5
         claude-fable-5-1: anthropic/claude-fable-5-1
         claude-fable-5: anthropic/claude-fable-5
         claude-opus-4-8: anthropic/claude-opus-4-8

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-gemini-flash-local.yaml

@@ -1,13 +1,11 @@
 # Template: litellm-gemini-flash-local
-# Gemini 3.7 Flash via local LiteLLM
+# Gemini 3.8 Flash via local LiteLLM
 #
 # Requires:
 #   - GEMINI_API_KEY in .env
 #
 # Backend (LiteLLM on port 4000) starts automatically.
 # All tiers use Flash, differentiated by reasoning_effort.
-# This route remains on 3.7 because LiteLLM 1.99's bundled model catalog has no
-# gemini-3.8-flash pricing/capability entry.
 #
 # After adding GEMINI_API_KEY, stop the backend so Forge restarts it with the new environment:
 # forge model backend stop litellm-4000
@@ -20,9 +18,16 @@ proxy:
 
   litellm:
     tiers:
-      haiku: gemini/gemini-3.7-flash
-      sonnet: gemini/gemini-3.7-flash
-      opus: gemini/gemini-3.7-flash
+      haiku: gemini/gemini-3.8-flash
+      sonnet: gemini/gemini-3.8-flash
+      opus: gemini/gemini-3.8-flash
+    model_alternatives:
+      haiku:
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      sonnet:
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      opus:
+        gemini-3.7-flash: gemini/gemini-3.7-flash
     # All tiers use the same model with different reasoning levels.
     tier_overrides:
       haiku:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-gemini-local.yaml

@@ -6,8 +6,6 @@
 #
 # Backend (LiteLLM on port 4000) starts automatically.
 # For tests: Use litellm-gemini-test template (port 4001)
-# Flash remains on 3.7 because LiteLLM 1.99's bundled model catalog has no
-# gemini-3.8-flash pricing/capability entry.
 
 proxy:
   family: gemini
@@ -17,9 +15,18 @@ proxy:
 
   litellm:
     tiers:
-      haiku: gemini/gemini-3.7-flash
+      haiku: gemini/gemini-3.8-flash
       sonnet: gemini/gemini-3.1-pro-preview
       opus: gemini/gemini-3.1-pro-preview
+    model_alternatives:
+      haiku:
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      sonnet:
+        gemini-3.8-flash: gemini/gemini-3.8-flash
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      opus:
+        gemini-3.8-flash: gemini/gemini-3.8-flash
+        gemini-3.7-flash: gemini/gemini-3.7-flash
     # Per-tier overrides to differentiate sonnet/opus (both use same model)
     tier_overrides:
       sonnet:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-gemini-test.yaml

@@ -17,9 +17,18 @@ proxy:
 
   litellm:
     tiers:
-      haiku: gemini/gemini-3.7-flash
+      haiku: gemini/gemini-3.8-flash
       sonnet: gemini/gemini-3.1-pro-preview
       opus: gemini/gemini-3.1-pro-preview
+    model_alternatives:
+      haiku:
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      sonnet:
+        gemini-3.8-flash: gemini/gemini-3.8-flash
+        gemini-3.7-flash: gemini/gemini-3.7-flash
+      opus:
+        gemini-3.8-flash: gemini/gemini-3.8-flash
+        gemini-3.7-flash: gemini/gemini-3.7-flash
     # Per-tier overrides to differentiate sonnet/opus (both use same model)
     tier_overrides:
       sonnet:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-gemini.yaml

@@ -1,9 +1,8 @@
 # Template: litellm-gemini
 # Gemini models via remote/shared LiteLLM
 #
-# Default template for Gemini access through a shared LiteLLM gateway. Flash
-# remains on 3.7 because LiteLLM 1.99's bundled model catalog has no
-# gemini-3.8-flash pricing/capability entry.
+# Default template for Gemini access through a shared LiteLLM gateway.
+# The shared instance must serve vertex_ai/gemini-3.8-flash.
 # Requires LITELLM_API_KEY and LITELLM_BASE_URL in .env (or environment).
 
 proxy:
@@ -14,7 +13,7 @@ proxy:
 
   litellm:
     tiers:
-      haiku: vertex_ai/gemini-3.7-flash
+      haiku: vertex_ai/gemini-3.8-flash
       sonnet: vertex_ai/gemini-3.1-pro-preview
       opus: vertex_ai/gemini-3.1-pro-preview
     model_alternatives:
@@ -24,11 +23,13 @@ proxy:
         gemini-3.6-flash: vertex_ai/gemini-3.6-flash
         gemini-3.7-flash: vertex_ai/gemini-3.7-flash
       sonnet:
+        gemini-3.8-flash: vertex_ai/gemini-3.8-flash
         gemini-2.5-flash: vertex_ai/gemini-2.5-flash
         gemini-3.5-flash: vertex_ai/gemini-3.5-flash
         gemini-3.6-flash: vertex_ai/gemini-3.6-flash
         gemini-3.7-flash: vertex_ai/gemini-3.7-flash
       opus:
+        gemini-3.8-flash: vertex_ai/gemini-3.8-flash
         gemini-2.5-flash: vertex_ai/gemini-2.5-flash
         gemini-3.5-flash: vertex_ai/gemini-3.5-flash
         gemini-3.6-flash: vertex_ai/gemini-3.6-flash

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-openai-codex-local.yaml

@@ -21,7 +21,14 @@ proxy:
       sonnet: openai/gpt-5.3-codex
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
+      sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
     tier_overrides:
       sonnet:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-openai-local.yaml

@@ -21,9 +21,15 @@ proxy:
       sonnet: openai/gpt-6-astra
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
       sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
     # Per-tier overrides to differentiate sonnet/opus (both use same model)
     tier_overrides:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/litellm-openai.yaml

@@ -1,7 +1,7 @@
 # Template: litellm-openai
 # OpenAI models via remote/shared LiteLLM
 #
-# Access to GPT-6 Astra and retained GPT/O-series models.
+# Access to GPT-6 Astra, Sol, Luna, and retained GPT/O-series models.
 # Requires LITELLM_API_KEY and LITELLM_BASE_URL in .env (or environment).
 #
 # Haiku uses gpt-5.4-mini; sonnet and opus both use gpt-6-astra,
@@ -19,9 +19,15 @@ proxy:
       sonnet: openai/gpt-6-astra
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
       sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-luna: openai/gpt-6-luna
         gpt-5.6-sol: openai/gpt-5.6-sol
     tier_overrides:
       sonnet:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/openrouter-anthropic.yaml

@@ -1,11 +1,11 @@
 # Template: openrouter-anthropic
 # Anthropic Claude models via OpenRouter (direct, no LiteLLM)
 #
-# Access to Claude Opus 5, Sonnet 5, Fable 5.1/5, Opus 4.8/4.6, Sonnet 4.6, and Haiku 4.5.
+# Access to Claude Opus 5.5/5, Sonnet 5, Fable 5.1/5, Opus 4.8/4.6, Sonnet 4.6, and Haiku 4.5.
 # Requires OPENROUTER_API_KEY in .env (or environment).
 #
-# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5. Alternatives are
-# claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
+# Defaults: sonnet tier -> Sonnet 5, opus tier -> Opus 5.5. Alternatives are
+# claude-opus-5, claude-fable-5-1, claude-fable-5, claude-opus-4-8, claude-opus-4-6,
 # and claude-sonnet-4-6.
 
 proxy:
@@ -19,12 +19,13 @@ proxy:
     tiers:
       haiku: anthropic/claude-haiku-4.5
       sonnet: anthropic/claude-sonnet-5
-      opus: anthropic/claude-opus-5
+      opus: anthropic/claude-opus-5.5
     zdr_fallbacks:
-      anthropic/claude-fable-5.1: anthropic/claude-opus-5
-      anthropic/claude-fable-5: anthropic/claude-opus-5
+      anthropic/claude-fable-5.1: anthropic/claude-opus-5.5
+      anthropic/claude-fable-5: anthropic/claude-opus-5.5
     model_alternatives:
       opus:
+        claude-opus-5: anthropic/claude-opus-5
         claude-fable-5-1: anthropic/claude-fable-5.1
         claude-fable-5: anthropic/claude-fable-5
         claude-opus-4-8: anthropic/claude-opus-4.8

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/openrouter-deepseek.yaml

@@ -17,6 +17,15 @@ proxy:
       haiku: deepseek/deepseek-v4-flash
       sonnet: deepseek/deepseek-v4-pro
       opus: deepseek/deepseek-v4-pro
+    model_alternatives:
+      haiku:
+        deepseek-v4.1-flash: deepseek/deepseek-v4.1-flash
+      sonnet:
+        deepseek-v4.1-flash: deepseek/deepseek-v4.1-flash
+        deepseek-v4-pro-0813: deepseek/deepseek-v4-pro-0813
+      opus:
+        deepseek-v4.1-flash: deepseek/deepseek-v4.1-flash
+        deepseek-v4-pro-0813: deepseek/deepseek-v4-pro-0813
     tier_overrides:
       haiku:
         reasoning_effort: high

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/openrouter-glm.yaml

@@ -16,6 +16,16 @@ proxy:
       haiku: z-ai/glm-4.7-flash
       sonnet: z-ai/glm-5.3
       opus: z-ai/glm-5.3
+    model_alternatives:
+      haiku:
+        glm-5.3-flash: z-ai/glm-5.3-flash
+        glm-5.3-flashx: z-ai/glm-5.3-flashx
+      sonnet:
+        glm-5.3-flash: z-ai/glm-5.3-flash
+        glm-5.3-flashx: z-ai/glm-5.3-flashx
+      opus:
+        glm-5.3-flash: z-ai/glm-5.3-flash
+        glm-5.3-flashx: z-ai/glm-5.3-flashx
     tier_overrides:
       sonnet:
         reasoning_effort: high

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/openrouter-openai-codex.yaml

@@ -16,7 +16,19 @@ proxy:
       sonnet: openai/gpt-5.3-codex
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
+      sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-sol-pro: openai/gpt-6-sol-pro
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-sol-pro: openai/gpt-6-sol-pro
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
         gpt-5.6-sol: openai/gpt-5.6-sol
         gpt-6-astra-pro: openai/gpt-6-astra-pro
     tier_overrides:

{multi_forge-1.0.0 → multi_forge-1.0.1}/src/forge/config/defaults/templates/openrouter-openai.yaml

@@ -1,7 +1,7 @@
 # Template: openrouter-openai
 # OpenAI models via OpenRouter (direct, no LiteLLM)
 #
-# Access to GPT-6 Astra and GPT-5.4-mini.
+# Access to GPT-6 Astra, Sol, Luna, their Pro variants, and GPT-5.4-mini.
 # Requires OPENROUTER_API_KEY in .env (or environment).
 #
 # Haiku uses gpt-5.4-mini; sonnet and opus both use gpt-6-astra,
@@ -20,10 +20,21 @@ proxy:
       sonnet: openai/gpt-6-astra
       opus: openai/gpt-6-astra
     model_alternatives:
+      haiku:
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
       sonnet:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-sol-pro: openai/gpt-6-sol-pro
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
         gpt-5.6-sol: openai/gpt-5.6-sol
         gpt-6-astra-pro: openai/gpt-6-astra-pro
       opus:
+        gpt-6-sol: openai/gpt-6-sol
+        gpt-6-sol-pro: openai/gpt-6-sol-pro
+        gpt-6-luna: openai/gpt-6-luna
+        gpt-6-luna-pro: openai/gpt-6-luna-pro
         gpt-5.6-sol: openai/gpt-5.6-sol
         gpt-6-astra-pro: openai/gpt-6-astra-pro
     tier_overrides: