lambda-erp 1.1.6__tar.gz → 1.1.7__tar.gz

{lambda_erp-1.1.6 → lambda_erp-1.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: lambda-erp
-Version: 1.1.6
+Version: 1.1.7
 Summary: Core ERP logic - accounting, sales, purchasing, inventory
 Author: TORUS INVESTMENTS AG
 License-Expression: Apache-2.0

{lambda_erp-1.1.6 → lambda_erp-1.1.7}/api/chat.py

@@ -2860,14 +2860,14 @@ Users can attach PDFs and images (receipts, bills, contracts, screenshots) to th
 # ---------------------------------------------------------------------------
 # OpenAI report-spec specialist
 #
-# GPT-5.6 Terra stays as the planner/orchestrator. When it decides a report needs
+# GPT-6 Luna acts as the planner/orchestrator. When it decides a report needs
 # a custom analytics view it calls the create/update custom analytics tools.
-# Those handlers make a separate, narrowly prompted Terra call for the bounded
+# Those handlers make a separate, narrowly prompted GPT-6.1 Sol call for the bounded
 # declarative report spec. Both roles use the existing OPENAI_API_KEY.
 # ---------------------------------------------------------------------------
 
 
-REPORT_SPECIALIST_MODEL = "gpt-5.6-terra"
+REPORT_SPECIALIST_MODEL = "gpt-6.1-sol"
 
 
 def _report_specialist_model() -> str:
@@ -3276,17 +3276,17 @@ async def generate_title(
 # ---------------------------------------------------------------------------
 
 # The agentic orchestrator model. Function tools on /v1/chat/completions
-# require reasoning_effort="none" for the gpt-5.6 family (400 otherwise).
-ORCHESTRATOR_MODEL = "gpt-5.6-terra"
+# require reasoning_effort="none" for GPT-6 Luna (400 otherwise).
+ORCHESTRATOR_MODEL = "gpt-6-luna"
 
 
 # ---------------------------------------------------------------------------
-# LLM API selector: Chat Completions (default) vs Responses API
+# LLM API selector: Responses API (default) vs Chat Completions
 #
 # The Responses API (/v1/responses) is what accepts non-PDF file inputs
 # (Office documents via input_file), and it does reasoning natively (no
 # reasoning_effort="none" hack). Migrating the whole loop is risky, so it's
-# behind a flag and OFF by default. To keep the blast radius tiny, ALL
+# controlled by ERP_CHAT_API. To keep the blast radius tiny, ALL
 # conversation state stays Chat-Completions-shaped; we translate to/from the
 # Responses shape only at the single orchestrator call boundary
 # (_orchestrator_turn), and shim the Responses output back into the
@@ -3473,7 +3473,7 @@ def _orchestrator_turn(client, messages, tools, max_tokens):
         tools=tools,
         tool_choice="auto",
         max_completion_tokens=max_tokens,
-        # gpt-5.6-terra rejects function tools on /v1/chat/completions unless
+        # gpt-6-luna rejects function tools on /v1/chat/completions unless
         # reasoning_effort is "none" (400 otherwise).
         reasoning_effort="none",
     )
@@ -3490,10 +3490,11 @@ async def run_thinking_loop(
 ):
     """Run the agentic reasoning loop.
 
-    The orchestrator and report specialist both use OpenAI gpt-5.6-terra. When GPT decides to call
+    The orchestrator uses GPT-6 Luna; the report specialist uses GPT-6.1 Sol.
+    When GPT decides to call
     `create_custom_analytics_report` or `update_custom_analytics_report`
     with an intent/feedback hint, the tool handler itself delegates the
-    declarative-spec step to a separate narrowly prompted Terra call. We emit an
+    declarative-spec step to a separate narrowly prompted GPT-6.1 Sol call. We emit an
     `llm_provider` event around that delegation so the UI can surface it.
     """
     openai_api_key = os.environ.get("OPENAI_API_KEY", "")
@@ -3702,7 +3703,7 @@ async def run_thinking_loop(
             principal_ref[0] = refresh_auth_principal(principal_ref[0])
             live_role = principal_ref[0].get("role") if principal_ref[0] else None
 
-            # If the orchestrator is delegating report-spec generation to Terra,
+            # If the orchestrator is delegating report-spec generation to GPT-6.1 Sol,
             # surface the handoff in the UI.
             will_delegate_to_report_specialist = (
                 tool_allowed(fn_name, live_role)

{lambda_erp-1.1.6 → lambda_erp-1.1.7}/api/providers.py

@@ -1,8 +1,8 @@
 """Pricing data for external LLM APIs the chat orchestrator uses.
 
-Mirrors the shape of `lambda-web/backend/providers.py`. Only the models
-this app actually calls are listed — refresh this dict when model
-choices change. Prices are USD per 1M tokens.
+Mirrors the shape of `lambda-web/backend/providers.py`. Includes active models
+and GPT-6 alternatives for comparison. Prices are Standard USD per 1M tokens;
+see docs/llm-pricing.md for sources and billing scope.
 
 Consumers import `cost_of_openai_call` to turn an SDK `response.usage` object
 into a USD amount. Both Chat Completions and Responses usage shapes are
@@ -19,6 +19,36 @@ from typing import Any, Optional
 # ---------------------------------------------------------------------------
 
 OPENAI_PRICING: dict[str, dict[str, Any]] = {
+    # Official model pages, verified 2026-09-29. Above 272K input tokens,
+    # the higher rates apply to the entire request (including cache hits).
+    "gpt-6-astra": {
+        "tiered": True,
+        "tiers": [
+            {"max_input_tokens": 272_000, "input": 10.00, "cached_input": 1.00, "cache_write": 12.50, "output": 50.00},
+            {"max_input_tokens": None,    "input": 20.00, "cached_input": 2.00, "cache_write": 25.00, "output": 75.00},
+        ],
+    },
+    "gpt-6-sol": {
+        "tiered": True,
+        "tiers": [
+            {"max_input_tokens": 272_000, "input": 2.00, "cached_input": 0.20, "cache_write": 2.50, "output": 10.00},
+            {"max_input_tokens": None,    "input": 4.00, "cached_input": 0.40, "cache_write": 5.00, "output": 15.00},
+        ],
+    },
+    "gpt-6-luna": {
+        "tiered": True,
+        "tiers": [
+            {"max_input_tokens": 272_000, "input": 0.10, "cached_input": 0.01, "cache_write": 0.125, "output": 0.50},
+            {"max_input_tokens": None,    "input": 0.20, "cached_input": 0.02, "cache_write": 0.25, "output": 0.75},
+        ],
+    },
+    "gpt-6.1-sol": {
+        "tiered": True,
+        "tiers": [
+            {"max_input_tokens": 272_000, "input": 2.00, "cached_input": 0.10, "cache_write": 2.50, "output": 10.00},
+            {"max_input_tokens": None,    "input": 4.00, "cached_input": 0.20, "cache_write": 5.00, "output": 15.00},
+        ],
+    },
     "gpt-5.4": {
         "tiered": True,
         "tiers": [
@@ -31,8 +61,8 @@ OPENAI_PRICING: dict[str, dict[str, Any]] = {
     "gpt-5.6-terra": {
         "tiered": True,
         "tiers": [
-            {"max_input_tokens": 272_000, "input": 2.50, "cached_input": 0.25, "cache_write": 3.125, "output": 15.00},
-            {"max_input_tokens": None,    "input": 5.00, "cached_input": 0.50, "cache_write": 6.25,  "output": 22.50},
+            {"max_input_tokens": 272_000, "input": 2.00, "cached_input": 0.20, "cache_write": 2.50, "output": 12.00},
+            {"max_input_tokens": None,    "input": 4.00, "cached_input": 0.40, "cache_write": 5.00, "output": 18.00},
         ],
     },
     "gpt-4.1-nano": {"input": 0.10, "cached_input": 0.025, "output": 0.40},
@@ -54,8 +84,8 @@ TRANSCRIBE_PRICING: dict[str, dict[str, float]] = {
 
 
 # ---------------------------------------------------------------------------
-# Rate lookup — unknown models fall back to the most expensive model in each
-# family so we never under-count a call we forgot to register.
+# Rate lookup — preserve the legacy gpt-5.4 estimate for unknown model IDs.
+# Register a model explicitly before using it; this fallback is not a price cap.
 # ---------------------------------------------------------------------------
 
 def get_openai_rates(model: str, input_tokens: int = 0) -> dict[str, float]:

{lambda_erp-1.1.6 → lambda_erp-1.1.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "lambda-erp"
-version = "1.1.6"
+version = "1.1.7"
 description = "Core ERP logic - accounting, sales, purchasing, inventory"
 readme = "README.md"
 license = "Apache-2.0"