Skip to content

Automate browser based workflows with AI

GPT-4.1-support (#2159)

Suchintanc829966936f9
diff --git a/setup.sh b/setup.sh
index be98c53ac..48c50e63d 100755
--- a/setup.sh
+++ b/setup.sh
@@ -52,7 +52,12 @@ setup_llm_providers() {
else
update_or_add_env_var "OPENAI_API_KEY" "$openai_api_key"
update_or_add_env_var "ENABLE_OPENAI" "true"
- model_options+=("OPENAI_GPT4_TURBO" "OPENAI_GPT4V" "OPENAI_GPT4O")
+ model_options+=(
+ "OPENAI_GPT4_1"
+ "OPENAI_GPT4_1_MINI"
+ "OPENAI_GPT4_1_NANO"
+ "OPENAI_GPT4O"
+ )
fi
else
update_or_add_env_var "ENABLE_OPENAI" "false"
@@ -69,7 +74,10 @@ setup_llm_providers() {
else
update_or_add_env_var "ANTHROPIC_API_KEY" "$anthropic_api_key"
update_or_add_env_var "ENABLE_ANTHROPIC" "true"
- model_options+=("ANTHROPIC_CLAUDE3_OPUS" "ANTHROPIC_CLAUDE3_SONNET" "ANTHROPIC_CLAUDE3_HAIKU" "ANTHROPIC_CLAUDE3.5_SONNET")
+ model_options+=(
+ "ANTHROPIC_CLAUDE3.5_SONNET"
+ "ANTHROPIC_CLAUDE3.7_SONNET"
+ )
fi
else
update_or_add_env_var "ENABLE_ANTHROPIC" "false"
@@ -92,7 +100,9 @@ setup_llm_providers() {
update_or_add_env_var "AZURE_API_BASE" "$azure_api_base"
update_or_add_env_var "AZURE_API_VERSION" "$azure_api_version"
update_or_add_env_var "ENABLE_AZURE" "true"
- model_options+=("AZURE_OPENAI_GPT4V")
+ model_options+=(
+ "AZURE_OPENAI_GPT4O"
+ )
fi
else
update_or_add_env_var "ENABLE_AZURE" "false"
@@ -112,7 +122,6 @@ setup_llm_providers() {
model_options+=(
"GEMINI_FLASH_2_0"
"GEMINI_FLASH_2_0_LITE"
- "GEMINI_PRO"
"GEMINI_2.5_PRO_PREVIEW_03_25"
"GEMINI_2.5_PRO_EXP_03_25"
)
diff --git a/skyvern/cli/commands.py b/skyvern/cli/commands.py
index d882e3980..463482465 100644
--- a/skyvern/cli/commands.py
+++ b/skyvern/cli/commands.py
@@ -221,7 +221,14 @@ def setup_llm_providers() -> None:
else:
update_or_add_env_var("OPENAI_API_KEY", openai_api_key)
update_or_add_env_var("ENABLE_OPENAI", "true")
- model_options.extend(["OPENAI_GPT4O"])
+ model_options.extend(
+ [
+ "OPENAI_GPT4_1",
+ "OPENAI_GPT4_1_MINI",
+ "OPENAI_GPT4_1_NANO",
+ "OPENAI_GPT4O",
+ ]
+ )
else:
update_or_add_env_var("ENABLE_OPENAI", "false")
@@ -238,8 +245,6 @@ def setup_llm_providers() -> None:
update_or_add_env_var("ENABLE_ANTHROPIC", "true")
model_options.extend(
[
- "ANTHROPIC_CLAUDE3_OPUS",
- "ANTHROPIC_CLAUDE3_HAIKU",
"ANTHROPIC_CLAUDE3.5_SONNET",
"ANTHROPIC_CLAUDE3.7_SONNET",
]
@@ -283,7 +288,6 @@ def setup_llm_providers() -> None:
[
"GEMINI_FLASH_2_0",
"GEMINI_FLASH_2_0_LITE",
- "GEMINI_PRO",
"GEMINI_2.5_PRO_PREVIEW_03_25",
"GEMINI_2.5_PRO_EXP_03_25",
]
@@ -856,5 +860,9 @@ def init() -> None:
print("\nMCP configuration is complete! Your AI applications are now ready to use Skyvern Cloud.")
if run_local:
+ print("\nInstalling Chromium browser...")
+ subprocess.run(["playwright", "install", "chromium"], check=True)
+ print("Chromium installation complete.")
+
print("\nTo start using Skyvern, run:")
print(" skyvern run server")
diff --git a/skyvern/config.py b/skyvern/config.py
index 223ab1537..992cf5c6b 100644
--- a/skyvern/config.py
+++ b/skyvern/config.py
@@ -99,7 +99,8 @@ class Settings(BaseSettings):
# LLM Configuration #
#####################
# ACTIVE LLM PROVIDER
- LLM_KEY: str = "OPENAI_GPT4O"
+ LLM_KEY: str = "OPENAI_GPT4O" # This is the model name
+ LLM_API_KEY: str | None = None # API key for the model
SECONDARY_LLM_KEY: str | None = None
SELECT_AGENT_LLM_KEY: str | None = None
SINGLE_CLICK_AGENT_LLM_KEY: str | None = None
@@ -108,6 +109,8 @@ class Settings(BaseSettings):
LLM_CONFIG_TIMEOUT: int = 300
LLM_CONFIG_MAX_TOKENS: int = 4096
LLM_CONFIG_TEMPERATURE: float = 0
+ LLM_CONFIG_SUPPORT_VISION: bool = True # Whether the model supports vision
+ LLM_CONFIG_ADD_ASSISTANT_PREFIX: bool = False # Whether to add assistant prefix
# LLM PROVIDER SPECIFIC
ENABLE_OPENAI: bool = False
ENABLE_ANTHROPIC: bool = False
diff --git a/skyvern/forge/sdk/api/llm/config_registry.py b/skyvern/forge/sdk/api/llm/config_registry.py
index b03f1a3a3..500465d94 100644
--- a/skyvern/forge/sdk/api/llm/config_registry.py
+++ b/skyvern/forge/sdk/api/llm/config_registry.py
@@ -37,7 +37,17 @@ class LLMConfigRegistry:
@classmethod
def get_config(cls, llm_key: str) -> LLMRouterConfig | LLMConfig:
if llm_key not in cls._configs:
- raise InvalidLLMConfigError(llm_key)
+ # If the key is not found in registered configs, treat it as a general model
+ LOG.info("Using general model configuration for unknown LLM key", llm_key=llm_key)
+ if not llm_key:
+ raise InvalidLLMConfigError(f"LLM_KEY not set for {llm_key}")
+ return LLMConfig(
+ llm_key, # Use the LLM_KEY as the model name
+ ["LLM_API_KEY"],
+ supports_vision=settings.LLM_CONFIG_SUPPORT_VISION,
+ add_assistant_prefix=settings.LLM_CONFIG_ADD_ASSISTANT_PREFIX,
+ max_completion_tokens=settings.LLM_CONFIG_MAX_TOKENS,
+ )
return cls._configs[llm_key]
@@ -52,6 +62,36 @@ if settings.ENABLE_OPENAI:
add_assistant_prefix=False,
),
)
+ LLMConfigRegistry.register_config(
+ "OPENAI_GPT4_1",
+ LLMConfig(
+ "gpt-4.1",
+ ["OPENAI_API_KEY"],
+ supports_vision=True,
+ add_assistant_prefix=False,
+ max_completion_tokens=16384,
+ ),
+ )
+ LLMConfigRegistry.register_config(
+ "OPENAI_GPT4_1_MINI",
+ LLMConfig(
+ "gpt-4.1-mini",
+ ["OPENAI_API_KEY"],
+ supports_vision=True,
+ add_assistant_prefix=False,
+ max_completion_tokens=16384,
+ ),
+ )
+ LLMConfigRegistry.register_config(
+ "OPENAI_GPT4_1_NANO",
+ LLMConfig(
+ "gpt-4.1-nano",
+ ["OPENAI_API_KEY"],
+ supports_vision=True,
+ add_assistant_prefix=False,
+ max_completion_tokens=16384,
+ ),
+ )
LLMConfigRegistry.register_config(
"OPENAI_GPT4_5",
LLMConfig(