janito 3.13.3__tar.gz → 3.14.1__tar.gz

{janito-3.13.3 → janito-3.14.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: janito
-Version: 3.13.3
+Version: 3.14.1
 Summary: A new Python package called janito.
 Author-email: João Pinto <janito@ikignosis.org>
 Project-URL: Homepage, https://github.com/ikignosis/janito

{janito-3.13.3 → janito-3.14.1}/docs/configuration.md

@@ -101,4 +101,26 @@ You can override the default model using `--model MODEL_NAME` on any command:
 janito chat --provider openai --model gpt-4o
 ```
 
+### Model@Provider Syntax
+
+For convenience, you can specify both model and provider in a single argument using the `model@provider` syntax:
+
+```bash
+# These are equivalent:
+janito chat --provider openai --model gpt-4o
+janito chat -m gpt-4o@openai
+
+# More examples:
+janito -m claude-3-5-sonnet-20241022@anthropic "Write a Python function"
+janito -m kimi-k1-8k@moonshot "Translate this to Chinese"
+janito -m deepseek-chat@deepseek "Explain machine learning"
+```
+
+This syntax is particularly useful for:
+- Quick one-off commands
+- Scripts and automation
+- Following conventions from tools like Docker
+
+> **Note**: If you specify both `-m model@provider` and `-p provider`, the explicit `-p` flag takes precedence.
+
 > **Note**: The list of available models is automatically synchronized with the codebase. Use `janito list-models --provider PROVIDER` to see all available options.

{janito-3.13.3 → janito-3.14.1}/docs/guides/single-shot-terminal.md

@@ -10,6 +10,21 @@ janito "what are the key classes of this project?"
 
 Janito will process your request and exit after displaying the result.
 
+### Model and Provider Selection
+
+You can specify which model and provider to use for your one-shot command:
+
+```bash
+# Traditional approach
+janito -p openai -m gpt-4 "what are the key classes of this project?"
+
+# Shorthand syntax (model@provider)
+janito -m gpt-4@openai "what are the key classes of this project?"
+
+# Using a different provider
+janito -m claude-3-5-sonnet-20241022@anthropic "analyze this codebase"
+```
+
 ### Shell Command Execution
 
 Janito supports executing shell commands in one-shot mode using the built-in `run_bash_command` and `run_powershell_command` tools:

{janito-3.13.3 → janito-3.14.1}/docs/guides/using.md

@@ -23,6 +23,22 @@ janito
 - Use CLI flags to customize behavior (see [CLI Options](../reference/cli-options.md)).
 - Shell commands can be executed using the built-in `run_bash_command` and `run_powershell_command` tools (when enabled).
 
+### Model and Provider Selection
+
+You can specify which model and provider to use in two ways:
+
+**Traditional approach:**
+```bash
+janito -p openai -m gpt-4 "Your prompt here"
+```
+
+**Shorthand syntax:**
+```bash
+janito -m gpt-4@openai "Your prompt here"
+```
+
+The `model@provider` syntax is particularly useful for quick one-off commands and follows conventions used by tools like Docker.
+
 ## More Resources
 
 - [How Janito Uses Tools](using_tools.md): Automatic tool selection details.

{janito-3.13.3 → janito-3.14.1}/docs/index.md

@@ -33,6 +33,22 @@ Janito is a flexible and extensible platform for interacting with large language
    janito chat
    ```
 
+## Quick Examples
+
+```bash
+# Simple prompt
+janito "Create a Python script to calculate fibonacci numbers"
+
+# Using specific model and provider
+janito -p openai -m gpt-4 "Explain quantum computing"
+
+# Using the shorthand model@provider syntax
+janito -m gpt-4@openai "Explain quantum computing"
+
+# Interactive mode
+janito --interactive
+```
+
 ## Documentation
 
 Explore our comprehensive documentation:

{janito-3.13.3 → janito-3.14.1}/docs/reference/cli-options.md

@@ -36,7 +36,7 @@ These options are useful for one-off runs, scripting, or experimentation. They t
 | `-s SYSTEM_PROMPT`, `--system SYSTEM_PROMPT` | Set a system prompt |
 | `-S`, `--show-system` | Show the resolved system prompt for the main agent |
 | `-p PROVIDER`, `--provider PROVIDER` | Select the provider |
-| `-m MODEL`, `--model MODEL` | Select the model |
+| `-m MODEL`, `--model MODEL` | Select the model (supports `model@provider` syntax) |
 | `-t TEMPERATURE`, `--temperature TEMPERATURE` | Set the temperature |
 | `-v`, `--verbose` | Print extra information before answering |
 | `-R`, `--raw` | Print the raw JSON response from the OpenAI API (if applicable) |
@@ -49,6 +49,7 @@ These options are useful for one-off runs, scripting, or experimentation. They t
 ```sh
 janito [options] [prompt]
 janito -p openai -m gpt-3.5-turbo "Your prompt here"
+janito -m gpt-4@openai "Your prompt here"  # Using model@provider syntax
 janito -c myproject -p openai "Prompt for my project (uses ~/.janito/configs/myproject.json)"
 janito --list-tools
 janito --multi  # Start chat mode with multiline input as default
@@ -96,6 +97,44 @@ janito -u "Do something with C:/Windows/System32/hosts"
 
 _This page is generated from the output of `janito --help`._
 
+## 🔄 Model@Provider Syntax
+
+The `-m`/`--model` option supports a convenient `model@provider` syntax that allows you to specify both the model and provider in a single argument:
+
+```sh
+# Traditional approach (two arguments)
+janito -p openai -m gpt-4 "Your prompt"
+
+# New syntax (single argument) - equivalent to above
+janito -m gpt-4@openai "Your prompt"
+```
+
+### Benefits
+- **Shorter commands**: Combine model and provider selection
+- **Consistent with other tools**: Familiar syntax used by tools like Docker
+- **Backward compatible**: Existing `-p provider -m model` syntax still works
+
+### Examples
+
+```sh
+# Use GPT-4 with OpenAI
+janito -m gpt-4@openai "Explain quantum computing"
+
+# Use Claude with Anthropic
+janito -m claude-3-5-sonnet-20241022@anthropic "Write a Python function"
+
+# Use Kimi with Moonshot
+janito -m kimi-k1-8k@moonshot "Translate this to Chinese"
+
+# Complex model names work too
+janito -m gpt-4-turbo-preview@openai "Debug this code"
+```
+
+### Notes
+- If you specify both `-m model@provider` and `-p provider`, the explicit `-p` flag takes precedence
+- Multiple `@` symbols are handled by splitting on the last `@` (e.g., `model@with@symbols@provider` → model: `model@with@symbols`, provider: `provider`)
+- Empty parts are ignored (e.g., `@provider` or `model@` won't set the provider)
+
 ## 🧠 About `--effort`
 
 The `--effort` option allows you to set the reasoning effort for models that support it. This can influence how much computational or logical effort the model applies to your prompt. The available values are:

{janito-3.13.3 → janito-3.14.1}/docs/supported-providers-models.md

@@ -1,135 +1,137 @@
-# Supported Providers and Models
-
-This page lists the supported providers and their available models, organized by openness and sorted alphabetically within each category.
-
-> 📋 **Platform & Documentation Access**: For information about which providers have publicly accessible platforms and documentation vs. blocked access, see [Platform & Documentation Access](provider-platform-access.md).
-
-## Open-Source / Open-Weight Models
-
-### Alibaba
-
-- **Qwen3 235B A22B Instruct** (default) - Latest 1M context model
-- **Qwen3 235B A22B Thinking** - Reasoning-focused version
-- **Qwen3 30B A3B Instruct** - Compact 1M context model
-- **Qwen3 30B A3B Thinking** - Compact reasoning version
-- **Qwen3 Coder Plus** - Specialized for programming tasks
-- **Qwen3 Coder 480B** - Large-scale coding model
-- **Qwen Turbo** - High-speed general purpose
-- **Qwen Plus** - Balanced performance
-- **Qwen Max** - Maximum capability
-
-### Cerebras
-
-**Production Models:**
-
-- **Qwen-3 32B** - General instruction following
-
-**Preview Models:**
-
-- **Qwen-3 Coder 480B** - Programming-focused with 32k context
-- **Qwen-3 235B A22B Instruct** - Large-scale instruction model
-- **Qwen-3 235B A22B Thinking** - Reasoning-focused version
-- **GPT-OSS 120B** - Open-source model
-
-**Notes:**
-
-- All Cerebras models support 128k context window
-- Models are optimized for low-latency inference
-- Pricing varies by model size and capability
-
-### DeepSeek
-
-- **DeepSeek Chat** (default) - General purpose chat model (8K context)
-- **DeepSeek Reasoner** - Specialized for complex reasoning tasks (8K context)
-- **DeepSeek V3.1** - Latest V3.1 model with 128K context window
-- **DeepSeek V3.1 Base** - Base V3.1 model with 128K context window
-- **DeepSeek R1** - R1 model with 128K context window (reasoning-focused)
-
-### Mistral
-
-- **Mistral Large Latest** (default) - Most capable Mistral model with 128k context
-- **Mistral Medium Latest** - Balanced performance with 32k context
-- **Mistral Small Latest** - Compact and efficient with 32k context
-- **Codestral Latest** - Specialized for code generation with 256k context
-- **Codestral 2405** - Previous version of code-focused model
-- **Devstral Small Latest** - Optimized for agentic tool use in software development
-- **Devstral Medium Latest** - Enhanced agentic capabilities for development tasks
-
-Mistral provides both general-purpose and specialized models, with Codestral specifically designed for code generation and Devstral for agentic software development.
-
-For setup instructions, see the [Mistral Setup Guide](mistral-setup.md).
-
-### Moonshot
-
-- **Kimi K2 0905 Preview** (default) - Latest generation with enhanced performance
-- **Kimi K2 Turbo Preview** - Turbo version with optimized speed
-- **Kimi K2 0711 Preview** - Previous preview version
-
-Moonshot provides open-source Kimi models with competitive performance.
-
-### Z.AI
-
-- **GLM-4.5** (default) - Advanced reasoning and conversation
-- **GLM-4.5 Air** - Compact and efficient version
-
-### IBM WatsonX
-
-**Open-Source Models:**
-
-- **openai/gpt-oss-120b** (default) - Open-source 120B model with thinking capabilities
-- **openai/gpt-oss-20b** - Open-source 20B model with thinking capabilities
-
-**IBM Granite Models:**
-
-- **ibm/granite-3-8b-instruct** - IBM's Granite 3 8B Instruct model with 128K context
-- **ibm/granite-3-3-8b-instruct** - Updated Granite 3.3 8B Instruct model
-
-**Hosted Models:**
-
-- **meta-llama/llama-3-1-70b-instruct** - Meta Llama 3.1 70B hosted on WatsonX
-- **meta-llama/llama-3-3-70b-instruct** - Meta Llama 3.3 70B hosted on WatsonX
-- **mistralai/mistral-large** - Mistral Large model hosted on WatsonX
-- **mistralai/mistral-large-2407** - Mistral Large 2407 version
-
-IBM WatsonX provides access to IBM's Granite models as well as popular open-source models hosted on their platform. All models support 128K context windows.
-
-## Proprietary Models
-
-### Anthropic
-
-- **Claude 3.7 Sonnet** (default) - Latest Claude model with enhanced reasoning
-- **Claude 4 Opus** - Most capable Claude model (preview)
-- **Claude 4 Sonnet** - Balanced performance and capability (preview)
-- **Claude 3.5 Sonnet** - Previous generation, still highly capable
-- **Claude 3.5 Haiku** - Fast and efficient
-- **Claude 3 Haiku** - Compact and cost-effective
-
-### Google
-
-- **Gemini 2.5 Flash** (default) - Fast and efficient
-- **Gemini 2.5 Pro** - Advanced capabilities with extended context
-- **Gemini 2.5 Flash-Lite Preview** - Lightweight preview version
-
-### OpenAI
-
-- **GPT-5** (default) - Latest generation with advanced reasoning
-- **GPT-5 Mini** - Compact version of GPT-5
-- **GPT-5 Nano** - Ultra-compact version
-- **GPT-4.1** - Enhanced GPT-4 with improved capabilities
-- **GPT-4.1-mini** - Balanced performance and efficiency
-- **GPT-4.1-nano** - Lightweight version
-- **GPT-4 Turbo** - High-performance GPT-4 variant
-- **GPT-4o** - Multimodal GPT-4 optimized for chat
-- **GPT-4o-mini** - Compact multimodal version
-- **o3** - Advanced reasoning model
-- **o3-mini** - Compact reasoning model
-- **o4-mini** - Latest mini reasoning model
-- **GPT-3.5 Turbo** - Cost-effective general purpose
-
-For setup instructions, see the [OpenAI Setup Guide](openai-setup.md).
-
-### Azure OpenAI
-
-- **Azure OpenAI Deployment** - Custom Azure OpenAI deployments
-
-For setup instructions, see [Using Azure OpenAI with Janito](reference/azure-openai.md).
+# Supported Providers and Models
+
+> **💡 Quick Tip**: Use the convenient `model@provider` syntax to specify both model and provider in one argument: `janito -m gpt-4@openai "Your prompt"`
+
+This page lists the supported providers and their available models, organized by openness and sorted alphabetically within each category.
+
+> 📋 **Platform & Documentation Access**: For information about which providers have publicly accessible platforms and documentation vs. blocked access, see [Platform & Documentation Access](provider-platform-access.md).
+
+## Open-Source / Open-Weight Models
+
+### Alibaba
+
+- **Qwen3 235B A22B Instruct** (default) - Latest 1M context model
+- **Qwen3 235B A22B Thinking** - Reasoning-focused version
+- **Qwen3 30B A3B Instruct** - Compact 1M context model
+- **Qwen3 30B A3B Thinking** - Compact reasoning version
+- **Qwen3 Coder Plus** - Specialized for programming tasks
+- **Qwen3 Coder 480B** - Large-scale coding model
+- **Qwen Turbo** - High-speed general purpose
+- **Qwen Plus** - Balanced performance
+- **Qwen Max** - Maximum capability
+
+### Cerebras
+
+**Production Models:**
+
+- **Qwen-3 32B** - General instruction following
+
+**Preview Models:**
+
+- **Qwen-3 Coder 480B** - Programming-focused with 32k context
+- **Qwen-3 235B A22B Instruct** - Large-scale instruction model
+- **Qwen-3 235B A22B Thinking** - Reasoning-focused version
+- **GPT-OSS 120B** - Open-source model
+
+**Notes:**
+
+- All Cerebras models support 128k context window
+- Models are optimized for low-latency inference
+- Pricing varies by model size and capability
+
+### DeepSeek
+
+- **DeepSeek Chat** (default) - General purpose chat model (8K context)
+- **DeepSeek Reasoner** - Specialized for complex reasoning tasks (8K context)
+- **DeepSeek V3.1** - Latest V3.1 model with 128K context window
+- **DeepSeek V3.1 Base** - Base V3.1 model with 128K context window
+- **DeepSeek R1** - R1 model with 128K context window (reasoning-focused)
+
+### Mistral
+
+- **Mistral Large Latest** (default) - Most capable Mistral model with 128k context
+- **Mistral Medium Latest** - Balanced performance with 32k context
+- **Mistral Small Latest** - Compact and efficient with 32k context
+- **Codestral Latest** - Specialized for code generation with 256k context
+- **Codestral 2405** - Previous version of code-focused model
+- **Devstral Small Latest** - Optimized for agentic tool use in software development
+- **Devstral Medium Latest** - Enhanced agentic capabilities for development tasks
+
+Mistral provides both general-purpose and specialized models, with Codestral specifically designed for code generation and Devstral for agentic software development.
+
+For setup instructions, see the [Mistral Setup Guide](mistral-setup.md).
+
+### Moonshot
+
+- **Kimi K2 0905 Preview** (default) - Latest generation with enhanced performance
+- **Kimi K2 Turbo Preview** - Turbo version with optimized speed
+- **Kimi K2 0711 Preview** - Previous preview version
+
+Moonshot provides open-source Kimi models with competitive performance.
+
+### Z.AI
+
+- **GLM-4.5** (default) - Advanced reasoning and conversation
+- **GLM-4.5 Air** - Compact and efficient version
+
+### IBM WatsonX
+
+**Open-Source Models:**
+
+- **openai/gpt-oss-120b** (default) - Open-source 120B model with thinking capabilities
+- **openai/gpt-oss-20b** - Open-source 20B model with thinking capabilities
+
+**IBM Granite Models:**
+
+- **ibm/granite-3-8b-instruct** - IBM's Granite 3 8B Instruct model with 128K context
+- **ibm/granite-3-3-8b-instruct** - Updated Granite 3.3 8B Instruct model
+
+**Hosted Models:**
+
+- **meta-llama/llama-3-1-70b-instruct** - Meta Llama 3.1 70B hosted on WatsonX
+- **meta-llama/llama-3-3-70b-instruct** - Meta Llama 3.3 70B hosted on WatsonX
+- **mistralai/mistral-large** - Mistral Large model hosted on WatsonX
+- **mistralai/mistral-large-2407** - Mistral Large 2407 version
+
+IBM WatsonX provides access to IBM's Granite models as well as popular open-source models hosted on their platform. All models support 128K context windows.
+
+## Proprietary Models
+
+### Anthropic
+
+- **Claude 3.7 Sonnet** (default) - Latest Claude model with enhanced reasoning
+- **Claude 4 Opus** - Most capable Claude model (preview)
+- **Claude 4 Sonnet** - Balanced performance and capability (preview)
+- **Claude 3.5 Sonnet** - Previous generation, still highly capable
+- **Claude 3.5 Haiku** - Fast and efficient
+- **Claude 3 Haiku** - Compact and cost-effective
+
+### Google
+
+- **Gemini 2.5 Flash** (default) - Fast and efficient
+- **Gemini 2.5 Pro** - Advanced capabilities with extended context
+- **Gemini 2.5 Flash-Lite Preview** - Lightweight preview version
+
+### OpenAI
+
+- **GPT-5** (default) - Latest generation with advanced reasoning
+- **GPT-5 Mini** - Compact version of GPT-5
+- **GPT-5 Nano** - Ultra-compact version
+- **GPT-4.1** - Enhanced GPT-4 with improved capabilities
+- **GPT-4.1-mini** - Balanced performance and efficiency
+- **GPT-4.1-nano** - Lightweight version
+- **GPT-4 Turbo** - High-performance GPT-4 variant
+- **GPT-4o** - Multimodal GPT-4 optimized for chat
+- **GPT-4o-mini** - Compact multimodal version
+- **o3** - Advanced reasoning model
+- **o3-mini** - Compact reasoning model
+- **o4-mini** - Latest mini reasoning model
+- **GPT-3.5 Turbo** - Cost-effective general purpose
+
+For setup instructions, see the [OpenAI Setup Guide](openai-setup.md).
+
+### Azure OpenAI
+
+- **Azure OpenAI Deployment** - Custom Azure OpenAI deployments
+
+For setup instructions, see [Using Azure OpenAI with Janito](reference/azure-openai.md).

{janito-3.13.3 → janito-3.14.1}/janito/cli/main_cli.py

@@ -173,7 +173,7 @@ definition = [
         },
     ),
     (["-p", "--provider"], {"metavar": "PROVIDER", "help": "Select the provider"}),
-    (["-m", "--model"], {"metavar": "MODEL", "help": "Select the model"}),
+    (["-m", "--model"], {"metavar": "MODEL", "help": "Select the model (can use model@provider syntax)"}),
     (
         ["-t", "--temperature"],
         {"type": float, "default": None, "help": "Set the temperature"},
@@ -304,11 +304,13 @@ class JanitoCLI:
 
         self.parser = argparse.ArgumentParser(
             description="Janito CLI - A tool for running LLM-powered workflows from the command line."
-            "\n\nExample usage: janito -p moonshot -m kimi-k1-8k 'Your prompt here'\n\n"
+            "\n\nExample usage: janito -p moonshot -m kimi-k1-8k 'Your prompt here'\n"
+            "Example usage: janito -m model@provider 'Your prompt here'\n\n"
             "Use -m or --model to set the model for the session.",
         )
         self._define_args()
         self.args = self.parser.parse_args()
+        self._parse_model_provider_syntax()
         self._set_all_arg_defaults()
         # Support custom config file via -c/--config
         if getattr(self.args, "config", None):
@@ -349,6 +351,18 @@ class JanitoCLI:
 
         self.rich_reporter = RichTerminalReporter(raw_mode=self.args.raw)
 
+    def _parse_model_provider_syntax(self):
+        """Parse -m model@provider syntax to split into model and provider."""
+        model = getattr(self.args, "model", None)
+        if model and "@" in model:
+            model_part, provider_part = model.rsplit("@", 1)
+            if model_part and provider_part:
+                # Only set provider if not already explicitly set via -p flag
+                if getattr(self.args, "provider", None) is None:
+                    self.args.provider = provider_part
+                # Always set the model part (without the @provider suffix)
+                self.args.model = model_part
+
     def _define_args(self):
         for argnames, argkwargs in definition:
             # Patch version argument dynamically with real version

{janito-3.13.3 → janito-3.14.1}/janito/docs/GETTING_STARTED.md

@@ -57,6 +57,9 @@ janito --set-config ibm project_id YOUR_PROJECT_ID
 
 # Verify it's working
 janito "Hello, can you introduce yourself?"
+
+# Or use the shorthand syntax
+janito -m ibm/granite-3-3-8b-instruct@ibm "Hello, can you introduce yourself?"
 ```
 
 ## Your First Commands
@@ -69,6 +72,10 @@ janito "Create a Python script to calculate fibonacci numbers"
 # With specific model
 janito -m kimi-k1-8k "Explain machine learning in simple terms"
 
+# Using model@provider syntax (shorthand)
+janito -m kimi-k1-8k@moonshot "Explain machine learning in simple terms"
+janito -m gpt-4@openai "Explain quantum computing"
+
 # Interactive chat mode
 janito --chat
 ```

{janito-3.13.3 → janito-3.14.1}/janito/providers/alibaba/model_info.py

@@ -133,4 +133,24 @@ MODEL_SPECS = {
         thinking=False,
         max_cot=65536,
     ),
+    "qwen-omni-turbo": LLMModelInfo(
+        name="qwen-omni-turbo",
+        context=32768,
+        max_response=2048,
+        category="Alibaba Qwen Omni Turbo Model (OpenAI-compatible)",
+        driver="OpenAIModelDriver",
+        thinking_supported=False,
+        thinking=False,
+        max_cot=2048,
+    ),
+    "qwen3-coder-flash": LLMModelInfo(
+        name="qwen3-coder-flash",
+        context=1000000,
+        max_response=65536,
+        category="Alibaba Qwen3 Coder Flash Model (OpenAI-compatible)",
+        driver="OpenAIModelDriver",
+        thinking_supported=True,
+        thinking=False,
+        max_cot=65536,
+    ),
 }

{janito-3.13.3 → janito-3.14.1}/janito.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: janito
-Version: 3.13.3
+Version: 3.14.1
 Summary: A new Python package called janito.
 Author-email: João Pinto <janito@ikignosis.org>
 Project-URL: Homepage, https://github.com/ikignosis/janito