Build AI methods with your coding agent, then run them anywhere: as an MCP for chatbots, as a webapp for people, or via API for your software.
Feature/more open source models (#42)
Louis Choquel87ab9ba5b9ae
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 3b34bf6c5..a4c4f9b65 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -13,7 +13,7 @@**Major Model Additions** - Claude 4 (Opus & Sonnet), Grok-3, and GPT-4 image generation are now in the house.### Pipeline Base Library update-- **New pipe** - `extract_page_contents_and_views_from_pdf` transfered from cookbook to base library (congrats on the promotion!). This pipe extracts text, linked images, **AND** page_view images (rendered pages) - it's very useful if you want to use Vision in follow-up pipes+- **New pipe** - `extract_page_contents_and_views_from_pdf` transferred from cookbook to base library (congrats on the promotion!). This pipe extracts text, linked images, **AND** page_view images (rendered pages) - it's very useful if you want to use Vision in follow-up pipes### Added@@ -56,7 +56,7 @@- Include `pyproject.toml` inside the project build.- Fix `ImggEngineFactory`: image generation (imgg) handle required format is `platform/model_name`-- pipelex cli: Added `list-pipes` method that can list all the availables pipes with descriptions.+- pipelex cli: Added `list-pipes` method that can list all the available pipes along with their descriptions.- Use a minimum version for `uv` instead of a fixed version- Implement `AGENTS.md` for Codex- Add tests for some of the `tools.misc`@@ -64,8 +64,8 @@## [v0.2.8] - 2025-05-28-- Defaults to `uv`-- Simplify llm provider config: All the API keys, urls, and regions are to be set in `.env`.+- Replaced `poetry` by `uv` for dependency management.+- Simplify llm provider config: All the API keys, urls, and regions now live in the `.env`.- Added logging level `OFF`, prevents any log from hitting the console## [v0.2.7] - 2025-05-26diff --git a/pipelex/cogt/llm/llm_models/llm_family.py b/pipelex/cogt/llm/llm_models/llm_family.pyindex b1e2d00eb..9c870f359 100644--- a/pipelex/cogt/llm/llm_models/llm_family.py+++ b/pipelex/cogt/llm/llm_models/llm_family.py@@ -13,6 +13,7 @@ class LLMCreator(StrEnum):MISTRAL = "Mistral"PERPLEXITY = "Perplexity"XAI = "XAI"+ ALIBABA = "Alibaba"@propertydef prompting_target(self) -> LLMPromptingTarget:@@ -25,7 +26,7 @@ class LLMCreator(StrEnum):return LLMPromptingTarget.MISTRALcase LLMCreator.GOOGLE:return LLMPromptingTarget.GEMINI- case LLMCreator.AMAZON | LLMCreator.PERPLEXITY | LLMCreator.META | LLMCreator.DEEPSEEK | LLMCreator.XAI:+ case LLMCreator.AMAZON | LLMCreator.PERPLEXITY | LLMCreator.META | LLMCreator.DEEPSEEK | LLMCreator.XAI | LLMCreator.ALIBABA:return LLMPromptingTarget.OPENAI@property@@ -49,6 +50,8 @@ class LLMCreator(StrEnum):return "red"case LLMCreator.XAI:return "green"+ case LLMCreator.ALIBABA:+ return "yellow"class LLMFamily(StrEnum):@@ -92,6 +95,8 @@ class LLMFamily(StrEnum):CUSTOM_LLAMA_4 = "custom-llama-4"CUSTOM_GEMMA_3 = "custom-gemma-3"+ CUSTOM_MISTRAL_SMALL_3_1 = "custom-mistral-small3.1"+ CUSTOM_QWEN_3 = "custom-qwen3"@propertydef creator(self) -> LLMCreator:@@ -112,6 +117,7 @@ class LLMFamily(StrEnum):| LLMFamily.MISTRAL_CODESTRAL| LLMFamily.MINISTRAL| LLMFamily.PIXTRAL+ | LLMFamily.CUSTOM_MISTRAL_SMALL_3_1):return LLMCreator.MISTRALcase LLMFamily.BEDROCK_MISTRAL_LARGE:@@ -132,6 +138,8 @@ class LLMFamily(StrEnum):return LLMCreator.XAIcase LLMFamily.CUSTOM_LLAMA_4 | LLMFamily.CUSTOM_GEMMA_3:return LLMCreator.META+ case LLMFamily.CUSTOM_QWEN_3:+ return LLMCreator.ALIBABA@propertydef prompting_target(self) -> LLMPromptingTarget:diff --git a/pipelex/cogt/llm/llm_models/llm_platform.py b/pipelex/cogt/llm/llm_models/llm_platform.pyindex 8557be0e4..e3b1548ba 100644--- a/pipelex/cogt/llm/llm_models/llm_platform.py+++ b/pipelex/cogt/llm/llm_models/llm_platform.py@@ -17,19 +17,19 @@ class LLMPlatform(StrEnum):MISTRAL = "mistral"OPENAI = "openai"PERPLEXITY = "perplexity"- VERTEXAI_OPENAI = "vertexai_openai"+ VERTEXAI = "vertexai"XAI = "xai"- CUSTOM_OPENAI = "custom_openai"+ CUSTOM_LLM = "custom_llm"@staticmethoddef list_openai_related() -> list["LLMPlatform"]:return [LLMPlatform.OPENAI,LLMPlatform.AZURE_OPENAI,- LLMPlatform.VERTEXAI_OPENAI,+ LLMPlatform.VERTEXAI,LLMPlatform.PERPLEXITY,LLMPlatform.XAI,- LLMPlatform.CUSTOM_OPENAI,+ LLMPlatform.CUSTOM_LLM,]@staticmethod@@ -46,13 +46,13 @@ class LLMPlatform(StrEnum):case (LLMPlatform.OPENAI| LLMPlatform.AZURE_OPENAI- | LLMPlatform.VERTEXAI_OPENAI+ | LLMPlatform.VERTEXAI| LLMPlatform.PERPLEXITY| LLMPlatform.ANTHROPIC| LLMPlatform.MISTRAL| LLMPlatform.BEDROCK_ANTHROPIC| LLMPlatform.XAI- | LLMPlatform.CUSTOM_OPENAI+ | LLMPlatform.CUSTOM_LLM):return Truecase LLMPlatform.BEDROCK:@@ -75,11 +75,11 @@ class LLMPlatform(StrEnum):return "spring_green2"case LLMPlatform.PERPLEXITY:return "gray"- case LLMPlatform.VERTEXAI_OPENAI:+ case LLMPlatform.VERTEXAI:return "deep_sky_blue"case LLMPlatform.XAI:return "green"- case LLMPlatform.CUSTOM_OPENAI:+ case LLMPlatform.CUSTOM_LLM:return "white"diff --git a/pipelex/cogt/llm/llm_worker_factory.py b/pipelex/cogt/llm/llm_worker_factory.pyindex d48a2eb44..38f70d400 100644--- a/pipelex/cogt/llm/llm_worker_factory.py+++ b/pipelex/cogt/llm/llm_worker_factory.py@@ -41,7 +41,7 @@ class LLMWorkerFactory:structure_method=structure_method,report_delegate=report_delegate,)- case LLMPlatform.VERTEXAI_OPENAI:+ case LLMPlatform.VERTEXAI:try:import google.auth # noqa: F401except ImportError as exc:@@ -61,7 +61,7 @@ class LLMWorkerFactory:structure_method=StructureMethod.INSTRUCTOR_VERTEX_JSON,report_delegate=report_delegate,)- case LLMPlatform.CUSTOM_OPENAI:+ case LLMPlatform.CUSTOM_LLM:from pipelex.plugins.openai.openai_factory import OpenAIFactoryfrom pipelex.plugins.openai.openai_llm_worker import OpenAILLMWorkerdiff --git a/pipelex/cogt/plugin_manager.py b/pipelex/cogt/plugin_manager.pyindex 2aa589064..85955cbf4 100644--- a/pipelex/cogt/plugin_manager.py+++ b/pipelex/cogt/plugin_manager.py@@ -9,55 +9,55 @@ from pipelex.types import StrEnumclass PluginHandle(StrEnum):- OPENAI_ASYNC = "openai_async"- AZURE_OPENAI_ASYNC = "azure_openai_async"- ANTHROPIC_ASYNC = "anthropic_async"- BEDROCK_ANTHROPIC_ASYNC = "bedrock_anthropic_async"- MISTRAL_ASYNC = "mistral_async"- BEDROCK_ASYNC = "bedrock_async"- PERPLEXITY_ASYNC = "perplexity_async"- VERTEXAI_OPENAI_ASYNC = "vertexai_openai_async"- XAI_OPENAI_ASYNC = "xai_openai_async"- CUSTOM_OPENAI_ASYNC = "custom_openai_async"- FAL_ASYNC = "fal_async"+ OPENAI_SDK = "openai_sdk"+ AZURE_OPENAI_SDK = "azure_openai_sdk"+ ANTHROPIC_SDK = "anthropic_sdk"+ BEDROCK_ANTHROPIC_SDK = "bedrock_anthropic_sdk"+ MISTRAL_SDK = "mistral_sdk"+ BEDROCK_SDK = "bedrock_sdk"+ PERPLEXITY_OPENAI_SDK = "perplexity_openai_sdk"+ VERTEXAI_OPENAI_SDK = "vertexai_openai_sdk"+ XAI_OPENAI_SDK = "xai_openai_sdk"+ CUSTOM_LLM_OPENAI_SDK = "custom_llm_openai_sdk"+ FAL_SDK = "fal_sdk"@staticmethoddef get_for_llm_platform(llm_platform: LLMPlatform) -> "PluginHandle":match llm_platform:case LLMPlatform.OPENAI:- return PluginHandle.OPENAI_ASYNC+ return PluginHandle.OPENAI_SDKcase LLMPlatform.AZURE_OPENAI:- return PluginHandle.AZURE_OPENAI_ASYNC+ return PluginHandle.AZURE_OPENAI_SDKcase LLMPlatform.ANTHROPIC:- return PluginHandle.ANTHROPIC_ASYNC+ return PluginHandle.ANTHROPIC_SDKcase LLMPlatform.MISTRAL:- return PluginHandle.MISTRAL_ASYNC+ return PluginHandle.MISTRAL_SDKcase LLMPlatform.BEDROCK:- return PluginHandle.BEDROCK_ASYNC+ return PluginHandle.BEDROCK_SDKcase LLMPlatform.BEDROCK_ANTHROPIC:- return PluginHandle.BEDROCK_ANTHROPIC_ASYNC+ return PluginHandle.BEDROCK_ANTHROPIC_SDKcase LLMPlatform.PERPLEXITY:- return PluginHandle.PERPLEXITY_ASYNC- case LLMPlatform.VERTEXAI_OPENAI:- return PluginHandle.VERTEXAI_OPENAI_ASYNC+ return PluginHandle.PERPLEXITY_OPENAI_SDK+ case LLMPlatform.VERTEXAI:+ return PluginHandle.VERTEXAI_OPENAI_SDKcase LLMPlatform.XAI:- return PluginHandle.XAI_OPENAI_ASYNC- case LLMPlatform.CUSTOM_OPENAI:- return PluginHandle.CUSTOM_OPENAI_ASYNC+ return PluginHandle.XAI_OPENAI_SDK+ case LLMPlatform.CUSTOM_LLM:+ return PluginHandle.CUSTOM_LLM_OPENAI_SDK@staticmethoddef get_for_ocr_engine(ocr_platform: OcrPlatform) -> "PluginHandle":match ocr_platform:case OcrPlatform.MISTRAL:- return PluginHandle.MISTRAL_ASYNC+ return PluginHandle.MISTRAL_SDK@staticmethoddef get_for_imgg_engine(imgg_platform: ImggPlatform) -> "PluginHandle":match imgg_platform:case ImggPlatform.FAL_AI:- return PluginHandle.FAL_ASYNC+ return PluginHandle.FAL_SDKcase ImggPlatform.OPENAI:- return PluginHandle.OPENAI_ASYNC+ return PluginHandle.OPENAI_SDKPluginManagerRoot = Dict[str, Any]diff --git a/pipelex/libraries/llm_integrations/custom.toml b/pipelex/libraries/llm_integrations/custom.tomlindex 9e07a072c..a551366be 100644--- a/pipelex/libraries/llm_integrations/custom.toml+++ b/pipelex/libraries/llm_integrations/custom.toml@@ -5,11 +5,25 @@ is_gen_object_supported = falseis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0, output = 0 }-platform_llm_id = { custom_openai = "gemma3:4b" }+platform_llm_id = { custom_llm = "gemma3:4b" }[custom-llama-4."llama4:scout".latest]is_gen_object_supported = falseis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0, output = 0 }-platform_llm_id = { custom_openai = "llama4:scout" }+platform_llm_id = { custom_llm = "llama4:scout" }++["custom-mistral-small3.1"."mistral-small3.1".latest]+is_gen_object_supported = false+is_vision_supported = true+max_prompt_images = 3000+cost_per_million_tokens_usd = { input = 0, output = 0 }+platform_llm_id = { custom_llm = "mistral-small3.1:24b" }++["custom-qwen3"."qwen3:8b".latest]+is_gen_object_supported = false+is_vision_supported = false+cost_per_million_tokens_usd = { input = 0, output = 0 }+platform_llm_id = { custom_llm = "qwen3:8b" }+# TODO: support <think> tokensdiff --git a/pipelex/libraries/llm_integrations/vertexai.toml b/pipelex/libraries/llm_integrations/vertexai.tomlindex c338f2d6a..f7cc08242 100644--- a/pipelex/libraries/llm_integrations/vertexai.toml+++ b/pipelex/libraries/llm_integrations/vertexai.toml@@ -5,39 +5,39 @@ is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0.075, output = 0.3 }-platform_llm_id = { vertexai_openai = "google/gemini-1.5-flash" }+platform_llm_id = { vertexai = "google/gemini-1.5-flash" }[gemini."gemini-1.5-pro".latest]is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 1.25, output = 5.0 }-platform_llm_id = { vertexai_openai = "google/gemini-1.5-pro" }+platform_llm_id = { vertexai = "google/gemini-1.5-pro" }[gemini."gemini-2.0-flash".latest]is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0.1, output = 0.4 }-platform_llm_id = { vertexai_openai = "google/gemini-2.0-flash" }+platform_llm_id = { vertexai = "google/gemini-2.0-flash" }[gemini."gemini-2.5-pro".latest]is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0.0, output = 0.0 }-platform_llm_id = { vertexai_openai = "google/gemini-2.5-pro-preview-05-06" }+platform_llm_id = { vertexai = "google/gemini-2.5-pro-preview-05-06" }[gemini."gemini-2.5-flash"."2025-04-17"]is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0.15, output = 0.6 }-platform_llm_id = { vertexai_openai = "google/gemini-2.5-flash-preview-04-17" }+platform_llm_id = { vertexai = "google/gemini-2.5-flash-preview-04-17" }[gemini."gemini-2.5-flash".latest]is_gen_object_supported = trueis_vision_supported = truemax_prompt_images = 3000cost_per_million_tokens_usd = { input = 0.15, output = 0.6 }-platform_llm_id = { vertexai_openai = "google/gemini-2.5-flash-preview-05-20" }+platform_llm_id = { vertexai = "google/gemini-2.5-flash-preview-05-20" }diff --git a/pipelex/plugins/openai/openai_factory.py b/pipelex/plugins/openai/openai_factory.pyindex 5fb4ae1d8..7cc9b6554 100644--- a/pipelex/plugins/openai/openai_factory.py+++ b/pipelex/plugins/openai/openai_factory.py@@ -53,7 +53,7 @@ class OpenAIFactory:openai_openai_config = get_config().plugins.openai_configapi_key = openai_openai_config.get_api_key(secrets_provider=get_secrets_provider())the_client = openai.AsyncOpenAI(api_key=api_key)- case LLMPlatform.VERTEXAI_OPENAI:+ case LLMPlatform.VERTEXAI:vertexai_config = get_config().plugins.vertexai_configendpoint, api_key = vertexai_config.configure(secrets_provider=get_secrets_provider())@@ -71,7 +71,7 @@ class OpenAIFactory:api_key=api_key,base_url=endpoint,)- case LLMPlatform.CUSTOM_OPENAI:+ case LLMPlatform.CUSTOM_LLM:custom_endpoint_config = get_config().plugins.custom_endpoint_configbase_url, api_key = custom_endpoint_config.configure(secrets_provider=get_secrets_provider())diff --git a/pipelex/plugins/openai/openai_llm_worker.py b/pipelex/plugins/openai/openai_llm_worker.pyindex 5d6c584a6..f027d3561 100644--- a/pipelex/plugins/openai/openai_llm_worker.py+++ b/pipelex/plugins/openai/openai_llm_worker.py@@ -84,6 +84,8 @@ class OpenAILLMWorker(LLMWorkerAbstract):| LLMFamily.GPT_4O| LLMFamily.CUSTOM_LLAMA_4| LLMFamily.CUSTOM_GEMMA_3+ | LLMFamily.CUSTOM_MISTRAL_SMALL_3_1+ | LLMFamily.CUSTOM_QWEN_3| LLMFamily.PERPLEXITY_SEARCH| LLMFamily.PERPLEXITY_RESEARCH| LLMFamily.PERPLEXITY_REASONING@@ -183,6 +185,8 @@ class OpenAILLMWorker(LLMWorkerAbstract):| LLMFamily.GPT_4O| LLMFamily.CUSTOM_LLAMA_4| LLMFamily.CUSTOM_GEMMA_3+ | LLMFamily.CUSTOM_MISTRAL_SMALL_3_1+ | LLMFamily.CUSTOM_QWEN_3| LLMFamily.PERPLEXITY_SEARCH| LLMFamily.PERPLEXITY_RESEARCH| LLMFamily.PERPLEXITY_REASONINGdiff --git a/pipelex/plugins/openai/openai_llms.py b/pipelex/plugins/openai/openai_llms.pyindex 4227709a5..93718486b 100644--- a/pipelex/plugins/openai/openai_llms.py+++ b/pipelex/plugins/openai/openai_llms.py@@ -10,7 +10,7 @@ from pipelex.plugins.openai.openai_factory import OpenAIFactoryasync def openai_list_available_models(llm_platform: LLMPlatform) -> List[Model]:openai_client_async = OpenAIFactory.make_openai_client(llm_platform=llm_platform)match llm_platform:- case LLMPlatform.VERTEXAI_OPENAI | LLMPlatform.PERPLEXITY:+ case LLMPlatform.VERTEXAI | LLMPlatform.PERPLEXITY:raise LLMSDKError(f"Platform '{llm_platform}' does not support listing models with OpenAI SDK")case _:passdiff --git a/pipelex/tools/templating/template_preprocessor.py b/pipelex/tools/templating/template_preprocessor.pyindex 1bef8981f..115e97859 100644--- a/pipelex/tools/templating/template_preprocessor.py+++ b/pipelex/tools/templating/template_preprocessor.py@@ -69,19 +69,19 @@ def preprocess_template(template: str) -> str:# TODO: allow escape patterns# Replace @?variable patterns (optional insertion) - must come before @variable- new_template = re.sub(r"@\?([a-zA-Z0-9_.]+)", replace_optional_at_variable, processed_template)+ new_template = re.sub(r"@\?(?![0-9])([a-zA-Z0-9_.]+)", replace_optional_at_variable, processed_template)if new_template != processed_template:changes_made = Trueprocessed_template = new_template# Replace @variable patterns- new_template = re.sub(r"@([a-zA-Z0-9_.]+)", replace_at_variable, processed_template)+ new_template = re.sub(r"@(?![0-9])([a-zA-Z0-9_.]+)", replace_at_variable, processed_template)if new_template != processed_template:changes_made = Trueprocessed_template = new_template# Replace $variable patterns- new_template = re.sub(r"\$([a-zA-Z0-9_.]+)", replace_dollar_variable, processed_template)+ new_template = re.sub(r"\$(?![0-9])([a-zA-Z0-9_.]+)", replace_dollar_variable, processed_template)if new_template != processed_template:changes_made = Trueprocessed_template = new_templatediff --git a/tests/pipelex/cogt/conftest.py b/tests/pipelex/cogt/conftest.pyindex 18959e70a..f5f68ca3c 100644--- a/tests/pipelex/cogt/conftest.py+++ b/tests/pipelex/cogt/conftest.py@@ -70,6 +70,8 @@ def llm_handle(request: pytest.FixtureRequest) -> str:# "gemini-2-pro",# "gemini-2-5-pro",# "gemini-2-5-flash",+ # "mistral-small3.1",+ # "qwen3:8b",])def llm_handle_for_vision(request: pytest.FixtureRequest) -> str:@@ -128,7 +130,7 @@ def llm_creator(request: pytest.FixtureRequest) -> LLMCreator:LLMPlatform.OPENAI,# LLMPlatform.PERPLEXITY,# LLMPlatform.VERTEXAI_OPENAI,- # LLMPlatform.CUSTOM_OPENAI,+ # LLMPlatform.CUSTOM_LLM,# LLMPlatform.XAI,])@@ -158,6 +160,8 @@ def llm_platform(request: pytest.FixtureRequest) -> LLMPlatform:# "sonar-pro",# "gemma3:4b",# "llama4:scout",+ # "mistral-small3.1:24b",+ # "qwen3:8b",])def llm_id(request: pytest.FixtureRequest) -> str:diff --git a/tests/tools/templating/test_template_preprocessor.py b/tests/tools/templating/test_template_preprocessor.pyindex ba72b0d44..16be6635c 100644--- a/tests/tools/templating/test_template_preprocessor.py+++ b/tests/tools/templating/test_template_preprocessor.py@@ -80,3 +80,45 @@ Optional notes:result = preprocess_template(template)expected = '{% if optional %}{{ optional|tag("optional") }}{% endif %} {{ required|tag("required") }}'assert result == expected++ def test_dollar_amounts_not_processed(self):+ """Test that dollar amounts are not processed as variables."""+ template = "The price is $10M and the budget is $1000.50"+ result = preprocess_template(template)+ assert result == template++ def test_mixed_dollar_amounts_and_variables(self):+ """Test mixing dollar amounts with dollar variables."""+ template = "The price is $10M and the budget is $budget_amount"+ result = preprocess_template(template)+ expected = "The price is $10M and the budget is {{ budget_amount|format() }}"+ assert result == expected++ def test_dollar_amounts_with_spaces(self):+ """Test dollar amounts with spaces after the dollar sign."""+ template = "The price is $ 10M and the budget is $ 1000.50"+ result = preprocess_template(template)+ assert result == template++ def test_at_with_numbers_not_processed(self):+ """Test that @ patterns followed by numbers are not processed."""+ template = "The version is @1.0 and the build is @2.3.4"+ result = preprocess_template(template)+ assert result == template++ def test_optional_at_with_numbers_not_processed(self):+ """Test that @? patterns followed by numbers are not processed."""+ template = "The version is @?1.0 and the build is @?2.3.4"+ result = preprocess_template(template)+ assert result == template++ def test_mixed_at_patterns_with_numbers(self):+ """Test mixing @ patterns with numbers and valid variables."""+ template = "Version @1.0, build @?2.3.4, and @valid_var with @?optional_var"+ result = preprocess_template(template)+ expected = (+ "Version @1.0, build @?2.3.4, and "+ '{{ valid_var|tag("valid_var") }} with '+ '{% if optional_var %}{{ optional_var|tag("optional_var") }}{% endif %}'+ )+ assert result == expected