Package com.tnsai.llm

Class ModelCapabilities

java.lang.Object

com.tnsai.llm.ModelCapabilities

All Implemented Interfaces:

LLMCapabilities

public class ModelCapabilities
extends Object
implements LLMCapabilities

Default implementation of LLMCapabilities with builder pattern.

ModelCapabilities provides a concrete implementation of the LLMCapabilities interface, making it easy to define model capabilities statically or dynamically.

Static Model Profiles

Pre-defined capability profiles for popular models:

 // Get predefined profile
 LLMCapabilities gpt4o = ModelCapabilities.GPT_4O;
 LLMCapabilities claude = ModelCapabilities.CLAUDE_SONNET_4;

 // Check capabilities
 if (gpt4o.supportsVision()) {
     // Use multimodal features
 }
 

Custom Capabilities

 LLMCapabilities custom = ModelCapabilities.builder()
     .providerName("CustomProvider")
     .modelId("my-model-v1")
     .supportsStreaming(true)
     .supportsVision(false)
     .supportsFunctionCalling(true)
     .supportsStructuredOutput(true)
     .maxInputTokens(32000)
     .maxOutputTokens(4096)
     .supportedModalities(Set.of(Modality.TEXT))
     .inputCostPer1KTokens(0.001)
     .outputCostPer1KTokens(0.002)
     .build();
 

See Also:

• LLMCapabilities
• LLMClient.getCapabilities()

Nested Class Summary

Nested classes/interfaces inherited from interface com.tnsai.llm.LLMCapabilities

LLMCapabilities.Modality

Field Summary

Fields

Modifier and Type	Field	Description
static final ModelCapabilities	CLAUDE_3_OPUS	Anthropic Claude 3 Opus capabilities
static final ModelCapabilities	CLAUDE_35_SONNET	Anthropic Claude 3.5 Sonnet capabilities
static final ModelCapabilities	CLAUDE_SONNET_4	Anthropic Claude Sonnet 4 capabilities
static final ModelCapabilities	CODEGEEX_4	ZhipuAI CodeGeeX-4 - Code specialized
static final ModelCapabilities	GEMINI_15_PRO	Google Gemini 1.5 Pro capabilities
static final ModelCapabilities	GEMINI_25_FLASH	Google Gemini 2.5 Flash capabilities
static final ModelCapabilities	GLM_4_FLASH	ZhipuAI GLM-4-Flash - Fast and efficient
static final ModelCapabilities	GLM_4_PLUS	ZhipuAI GLM-4-Plus - Most capable ZhipuAI model
static final ModelCapabilities	GLM_4V_PLUS	ZhipuAI GLM-4V-Plus - Vision capable model
static final ModelCapabilities	GPT_4_TURBO	OpenAI GPT-4-turbo capabilities
static final ModelCapabilities	GPT_4O	OpenAI GPT-4o capabilities
static final ModelCapabilities	GPT_4O_MINI	OpenAI GPT-4o-mini capabilities
static final ModelCapabilities	GROQ_LLAMA_33_70B	Groq Llama 3.3 70B capabilities
static final ModelCapabilities	LLAMA_32	Meta Llama 3.2 (via Ollama) capabilities
static final ModelCapabilities	MINIMAX_ABAB65S	MiniMax abab6.5s-chat - Balanced performance
static final ModelCapabilities	MINIMAX_ABAB65T	MiniMax abab6.5t-chat - Turbo model for fast responses
static final ModelCapabilities	MINIMAX_M2	MiniMax M2 - Flagship model with 1M context
static final ModelCapabilities	MINIMAX_M21	MiniMax M2.1 - Enhanced multilingual model
static final ModelCapabilities	MISTRAL_LARGE	Mistral Large capabilities
static final ModelCapabilities	O1	OpenAI o1 (reasoning) capabilities
static final ModelCapabilities	UNKNOWN	Unknown/default capabilities

Constructor Summary

Constructors

Constructor	Description
ModelCapabilities()

Method Summary

Modifier and Type	Method	Description
static ModelCapabilities	forModel(String modelName)	Returns the capabilities for a known model by name.
Map<String,Object>	getAdditionalCapabilities()	Returns additional provider-specific capabilities.
String	getDescription()	Returns a human-readable description of the model.
Optional<Long>	getEstimatedLatencyMs()	Estimated average latency for first token (TTFT) in milliseconds.
Optional<Double>	getInputCostPer1KTokens()	Cost per 1000 input tokens in USD.
int	getMaxInputTokens()	Maximum input tokens (context window).
int	getMaxOutputTokens()	Maximum output tokens the model can generate.
String	getModelId()	The model identifier.
Optional<String>	getModelVersion()	Model version if applicable.
Optional<Double>	getOutputCostPer1KTokens()	Cost per 1000 output tokens in USD.
String	getProviderName()	The provider name (OpenAI, Anthropic, Google, etc.).
Set<LLMCapabilities.Modality>	getSupportedModalities()	Returns the list of supported input modalities.
boolean	supportsCodeExecution()	Whether the model supports code execution (like Code Interpreter).
boolean	supportsFunctionCalling()	Whether the model supports function/tool calling.
boolean	supportsParallelFunctionCalls()	Whether the model supports parallel function calls.
boolean	supportsReasoning()	Whether the model has reasoning/thinking capabilities (like o1).
boolean	supportsStreaming()	Whether the model supports streaming responses.
boolean	supportsStructuredOutput()	Whether the model supports structured JSON output mode.
boolean	supportsSystemPrompt()	Whether the model supports system prompts.
boolean	supportsVision()	Whether the model can process images (vision capability).
boolean	supportsWebBrowsing()	Whether the model supports web browsing.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface com.tnsai.llm.LLMCapabilities

getContextWindow, knowsInputWindow, meetsRequirements, requireStreaming, requireToolCalling, requireVision, supportsFileUpload, supportsModality

Field Details

GPT_4O

public static final ModelCapabilities GPT_4O

OpenAI GPT-4o capabilities

GPT_4O_MINI

public static final ModelCapabilities GPT_4O_MINI

OpenAI GPT-4o-mini capabilities

GPT_4_TURBO

public static final ModelCapabilities GPT_4_TURBO

OpenAI GPT-4-turbo capabilities

O1

public static final ModelCapabilities O1

OpenAI o1 (reasoning) capabilities

CLAUDE_SONNET_4

public static final ModelCapabilities CLAUDE_SONNET_4

Anthropic Claude Sonnet 4 capabilities

CLAUDE_35_SONNET

public static final ModelCapabilities CLAUDE_35_SONNET

Anthropic Claude 3.5 Sonnet capabilities

CLAUDE_3_OPUS

public static final ModelCapabilities CLAUDE_3_OPUS

Anthropic Claude 3 Opus capabilities

GEMINI_25_FLASH

public static final ModelCapabilities GEMINI_25_FLASH

Google Gemini 2.5 Flash capabilities

GEMINI_15_PRO

public static final ModelCapabilities GEMINI_15_PRO

Google Gemini 1.5 Pro capabilities

LLAMA_32

public static final ModelCapabilities LLAMA_32

Meta Llama 3.2 (via Ollama) capabilities

MISTRAL_LARGE

public static final ModelCapabilities MISTRAL_LARGE

Mistral Large capabilities

GROQ_LLAMA_33_70B

public static final ModelCapabilities GROQ_LLAMA_33_70B

Groq Llama 3.3 70B capabilities

GLM_4_PLUS

public static final ModelCapabilities GLM_4_PLUS

ZhipuAI GLM-4-Plus - Most capable ZhipuAI model

GLM_4V_PLUS

public static final ModelCapabilities GLM_4V_PLUS

ZhipuAI GLM-4V-Plus - Vision capable model

GLM_4_FLASH

public static final ModelCapabilities GLM_4_FLASH

ZhipuAI GLM-4-Flash - Fast and efficient

CODEGEEX_4

public static final ModelCapabilities CODEGEEX_4

ZhipuAI CodeGeeX-4 - Code specialized

MINIMAX_M2

public static final ModelCapabilities MINIMAX_M2

MiniMax M2 - Flagship model with 1M context

MINIMAX_M21

public static final ModelCapabilities MINIMAX_M21

MiniMax M2.1 - Enhanced multilingual model

MINIMAX_ABAB65S

public static final ModelCapabilities MINIMAX_ABAB65S

MiniMax abab6.5s-chat - Balanced performance

MINIMAX_ABAB65T

public static final ModelCapabilities MINIMAX_ABAB65T

MiniMax abab6.5t-chat - Turbo model for fast responses

UNKNOWN

public static final ModelCapabilities UNKNOWN

Unknown/default capabilities

Constructor Details

ModelCapabilities

public ModelCapabilities()

Method Details

supportsStreaming

public boolean supportsStreaming()

Description copied from interface: LLMCapabilities

Whether the model supports streaming responses. Most modern LLMs support this.

Specified by:

supportsStreaming in interface LLMCapabilities

supportsVision

public boolean supportsVision()

Description copied from interface: LLMCapabilities

Whether the model can process images (vision capability). Examples: GPT-4o, Claude 3, Gemini

Specified by:

supportsVision in interface LLMCapabilities

supportsFunctionCalling

public boolean supportsFunctionCalling()

Description copied from interface: LLMCapabilities

Whether the model supports function/tool calling. This is critical for agent-based applications.

Specified by:

supportsFunctionCalling in interface LLMCapabilities

supportsStructuredOutput

public boolean supportsStructuredOutput()

Description copied from interface: LLMCapabilities

Whether the model supports structured JSON output mode. When enabled, model output is guaranteed to be valid JSON.

Specified by:

supportsStructuredOutput in interface LLMCapabilities

supportsSystemPrompt

public boolean supportsSystemPrompt()

Description copied from interface: LLMCapabilities

Whether the model supports system prompts. Some older models don't distinguish system from user messages.

Specified by:

supportsSystemPrompt in interface LLMCapabilities

supportsParallelFunctionCalls

public boolean supportsParallelFunctionCalls()

Description copied from interface: LLMCapabilities

Whether the model supports parallel function calls. Some models can request multiple tool calls in a single response.

Specified by:

supportsParallelFunctionCalls in interface LLMCapabilities

supportsCodeExecution

public boolean supportsCodeExecution()

Description copied from interface: LLMCapabilities

Whether the model supports code execution (like Code Interpreter).

Specified by:

supportsCodeExecution in interface LLMCapabilities

supportsWebBrowsing

public boolean supportsWebBrowsing()

Description copied from interface: LLMCapabilities

Whether the model supports web browsing.

Specified by:

supportsWebBrowsing in interface LLMCapabilities

supportsReasoning

public boolean supportsReasoning()

Description copied from interface: LLMCapabilities

Whether the model has reasoning/thinking capabilities (like o1).

Specified by:

supportsReasoning in interface LLMCapabilities

getMaxInputTokens

public int getMaxInputTokens()

Description copied from interface: LLMCapabilities

Maximum input tokens (context window). This is the combined limit for system prompt + history + user message.

Specified by:

getMaxInputTokens in interface LLMCapabilities

getMaxOutputTokens

public int getMaxOutputTokens()

Description copied from interface: LLMCapabilities

Maximum output tokens the model can generate. Some models have separate input/output limits.

Specified by:

getMaxOutputTokens in interface LLMCapabilities

getSupportedModalities

public Set<LLMCapabilities.Modality> getSupportedModalities()

Description copied from interface: LLMCapabilities

Returns the list of supported input modalities.

Specified by:

getSupportedModalities in interface LLMCapabilities

Returns:

Set of supported modalities (TEXT, IMAGE, AUDIO, etc.)

getProviderName

public String getProviderName()

Description copied from interface: LLMCapabilities

The provider name (OpenAI, Anthropic, Google, etc.).

Specified by:

getProviderName in interface LLMCapabilities

getModelId

public String getModelId()

Description copied from interface: LLMCapabilities

The model identifier.

Specified by:

getModelId in interface LLMCapabilities

getModelVersion

public Optional<String> getModelVersion()

Description copied from interface: LLMCapabilities

Model version if applicable.

Specified by:

getModelVersion in interface LLMCapabilities

getInputCostPer1KTokens

public Optional<Double> getInputCostPer1KTokens()

Description copied from interface: LLMCapabilities

Cost per 1000 input tokens in USD. Returns empty if pricing not available.

Specified by:

getInputCostPer1KTokens in interface LLMCapabilities

getOutputCostPer1KTokens

public Optional<Double> getOutputCostPer1KTokens()

Description copied from interface: LLMCapabilities

Cost per 1000 output tokens in USD. Returns empty if pricing not available.

Specified by:

getOutputCostPer1KTokens in interface LLMCapabilities

getEstimatedLatencyMs

public Optional<Long> getEstimatedLatencyMs()

Description copied from interface: LLMCapabilities

Estimated average latency for first token (TTFT) in milliseconds. Useful for latency-sensitive applications.

Specified by:

getEstimatedLatencyMs in interface LLMCapabilities

getAdditionalCapabilities

public Map<String,Object> getAdditionalCapabilities()

Description copied from interface: LLMCapabilities

Returns additional provider-specific capabilities. This is an extension point for capabilities not covered by the interface.

Specified by:

getAdditionalCapabilities in interface LLMCapabilities

getDescription

public String getDescription()

Description copied from interface: LLMCapabilities

Returns a human-readable description of the model.

Specified by:

getDescription in interface LLMCapabilities

forModel

public static ModelCapabilities forModel(String modelName)

Returns the capabilities for a known model by name.

Parameters:

modelName - The model name (e.g., "gpt-4o", "claude-sonnet-4")

Returns:

The capabilities, or UNKNOWN if not found