From f0f26086469b7c4b1fcbb4cc97388efd399a27c3 Mon Sep 17 00:00:00 2001
From: Salil Das <11658960+sadlilas@users.noreply.github.com>
Date: Sun, 22 Feb 2026 13:49:59 -0800
Subject: [PATCH] feat: add context_budget_cap and context window to provider
 defaults
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Gemini charges 2x input / 1.5x output above 200k tokens, matching
Anthropic's pricing cliff. Add context_budget_cap: 200,000 so
context-simple can cap its working budget at the cost-optimal threshold.

Also adds context_window (1,048,576) and max_output_tokens (65,536)
to ProviderInfo.defaults -- these were missing, causing context-simple
to fall through to its 200k fallback instead of calculating a dynamic
budget from the actual model limits.

Related: microsoft-amplifier/amplifier-support#57

🤖 Generated with [Amplifier](https://github.com/microsoft/amplifier)

Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>
---
 amplifier_module_provider_gemini/__init__.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/amplifier_module_provider_gemini/__init__.py b/amplifier_module_provider_gemini/__init__.py
index 479285c..2e149cc 100644
--- a/amplifier_module_provider_gemini/__init__.py
+++ b/amplifier_module_provider_gemini/__init__.py
@@ -200,6 +200,10 @@ def get_info(self) -> ProviderInfo:
                 "max_tokens": 8192,
                 "temperature": 0.7,
                 "timeout": 600.0,
+                "context_window": 1_048_576,
+                "max_output_tokens": 65_536,
+                # Gemini charges 2x input / 1.5x output above 200k tokens
+                "context_budget_cap": 200_000,
             },
             config_fields=[
                 ConfigField(