Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:257
Model provider for any API that speaks the OpenAI chat completions protocol:
OpenAI itself, and the compatible endpoints of DeepSeek, Qwen, Moonshot, Ollama,
vLLM and others. Set baseURL and model for the provider you use.
Ports:
prompt(in): a text prompt. The answer is sent oncompletion.messages(in): a list of chat messages. The reply is sent onmessage.
Extends
Component
Extended by
Constructors
Constructor
new OpenAICompatible(config): OpenAICompatible;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:268
Parameters
| Parameter | Type |
|---|---|
config |
OpenAICompatibleConfig |
Returns
OpenAICompatible
Overrides
Component.constructor
Properties
| Property | Modifier | Type | Description | Inherited from | Defined in |
|---|---|---|---|---|---|
flow |
public |
Flow | null |
- | Deepseek.flow |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:24 |
I |
public |
(name) => Port |
- | Component.I |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:26 |
id |
public |
string |
- | Component.id |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:18 |
identity |
public |
Symbol |
- | Component.identity |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:25 |
inPort |
protected readonly |
Port |
The default in input port. |
Component.inPort |
packages/core/dist/component/index.d.ts:49 |
isRunning |
public |
boolean |
- | Component.isRunning |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:23 |
O |
public |
(name) => Port |
- | Component.O |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:27 |
opts |
public |
object |
- | Component.opts |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:19 |
outPort |
protected readonly |
Port |
The default out output port. A pipeline uses it to infer the endpoint. |
Component.outPort |
packages/core/dist/component/index.d.ts:51 |
portMap |
public |
PortMap |
- | Component.portMap |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:22 |
type |
public |
Symbol |
- | Component.type |
node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:17 |
Port |
static |
typeof Port |
Port factory from the HLang runtime. Component.Port.I(name) and Component.Port.O(name) create input and output ports; call .attach(this) in the constructor. |
Component.Port |
packages/core/dist/component/index.d.ts:47 |
Methods
_transform()
_transform($i, $o): void;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:414
Connects the prompt and messages input ports to the model.
API errors produce a fallback value so the pipeline keeps flowing.
Parameters
| Parameter | Type |
|---|---|
$i |
Port |
$o |
Port |
Returns
void
Overrides
Component._transform
attach()
attach(flow): void;
Defined in: node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:31
Parameters
| Parameter | Type |
|---|---|
flow |
any |
Returns
void
Inherited from
Component.attach
attachPort()
attachPort(port): void;
Defined in: node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:32
Parameters
| Parameter | Type |
|---|---|
port |
any |
Returns
void
Inherited from
Component.attachPort
chatCompletion()
chatCompletion(messages, options?): Promise<DeepseekAssistantMessage>;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:317
Sends a conversation and returns the assistant reply.
Parameters
| Parameter | Type |
|---|---|
messages |
DeepseekMessage[] |
options? |
DeepseekChatCompletionOptions |
Returns
Promise<DeepseekAssistantMessage>
E()
E(): Port;
Defined in: node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:30
Returns
Port
Inherited from
Component.E
executeOnce()
executeOnce(type, ...args): void;
Defined in: node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/node.d.ts:29
Parameters
| Parameter | Type |
|---|---|
type |
any |
...args |
any[] |
Returns
void
Inherited from
Component.executeOnce
generateCompletion()
generateCompletion(prompt): Promise<string>;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:300
Generates a completion for a single prompt.
Parameters
| Parameter | Type |
|---|---|
prompt |
string |
Returns
Promise<string>
The text answer, or the JSON-encoded assistant message when it contains tool calls.
run()
run(input): Promise<
| string
| DeepseekAssistantMessage>;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:406
Runs the model outside a pipeline.
Parameters
| Parameter | Type | Description |
|---|---|---|
input |
string | DeepseekMessage[] |
A prompt string or a list of messages. |
Returns
Promise<
| string
| DeepseekAssistantMessage>
Overrides
Component.run
streamChatCompletion()
streamChatCompletion(messages, options?): AsyncGenerator<DeepseekStreamChunk>;
Defined in: packages/integrations/src/model-provider/openai-compatible/index.ts:359
Streams the assistant reply. Each chunk holds a text delta, tool-call deltas, or usage.
Parameters
| Parameter | Type |
|---|---|
messages |
DeepseekMessage[] |
options? |
DeepseekChatCompletionOptions |
Returns
AsyncGenerator<DeepseekStreamChunk>
throw()
throw(error): void;
Defined in: node_modules/.pnpm/@hlang-org+runtime@1.0.7/node_modules/@hlang-org/runtime/lib/node/transform.d.ts:4
Parameters
| Parameter | Type |
|---|---|
error |
any |
Returns
void
Inherited from
Component.throw