@agentskit/adapters — Variables

#Variable: azureOpenAIAdapter

const azureOpenAIAdapter: (config) => AdapterFactory = azureOpenAI

Defined in: adapters/src/azure-openai.ts:66

#Parameters

#config

AzureOpenAIConfig

#Returns

AdapterFactory

---

#Variable: bailAdapter

const bailAdapter: (config) => AdapterFactory = bail

Defined in: adapters/src/bail.ts:35

Alibaba Bailian (Qwen) via the DashScope OpenAI-compatibility endpoint.

Supports the Qwen-2.5 / 3 chat series and Qwen-VL multimodal models. APAC users typically prefer this over OpenAI for latency + data residency.

Default model: qwen-max.

#Parameters

#config

Partial<BailConfig> & object

#Returns

AdapterFactory

---

#Variable: bedrockAdapter

const bedrockAdapter: (config) => AdapterFactory = bedrock

Defined in: adapters/src/bedrock.ts:265

#Parameters

#config

BedrockConfig

#Returns

AdapterFactory

---

#Variable: cerebrasAdapter

const cerebrasAdapter: (config) => AdapterFactory = cerebras

Defined in: adapters/src/cerebras.ts:32

Cerebras — ultra-fast inference on wafer-scale chips. OpenAI-compatible endpoint serving Llama / Qwen models with very low first-token latency.

Default model: llama-3.3-70b.

#Parameters

#config

Partial<CerebrasConfig> & object

#Returns

AdapterFactory

---

#Variable: cohereAdapter

const cohereAdapter: (config) => AdapterFactory = cohere

Defined in: adapters/src/cohere.ts:39

Alias for naming consistency with other native adapters.

Cohere Command models via Cohere's OpenAI-compatibility endpoint.

• Streams tokens via SSE (OpenAI-compatible chunks).
• Supports tool calls in the OpenAI tools shape.
• Reports usage on the final stream chunk when the upstream model returns it (Cohere's compatibility layer mirrors OpenAI's stream_options: \{ include_usage: true \} semantics).
• Inherits auto-retry from the shared OpenAI core (retry).

Default model: command-r-plus. Override via model.

#Parameters

#config

Partial<CohereConfig> & object

#Returns

AdapterFactory

---

#Variable: deepseek

const deepseek: (config) => AdapterFactory

Defined in: adapters/src/deepseek.ts:5

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: deepseekEmbedder

const deepseekEmbedder: (config) => EmbedFn

Defined in: adapters/src/embedders/deepseek.ts:5

#Parameters

#config

OpenAICompatibleEmbedderConfig

#Returns

EmbedFn

---

#Variable: DEFAULT_CARBON_TABLE

const DEFAULT_CARBON_TABLE: CarbonTable

Defined in: adapters/src/carbon.ts:31

Default carbon table. Keys are provider:region; values are estimated gCO2eq per 1k tokens of typical mixed input/output. Lower is greener.

---

#Variable: DEFAULT_DEPRECATION_TABLE

const DEFAULT_DEPRECATION_TABLE: ModelDeprecation[]

Defined in: adapters/src/deprecation.ts:44

Known deprecations (current as of AgentsKit shipping). Bump on each release; PRs welcome. Successor picks lean towards the same provider's current default unless the entire family is gone.

---

#Variable: fireworks

const fireworks: (config) => AdapterFactory

Defined in: adapters/src/fireworks.ts:9

Fireworks AI. OpenAI-compatible endpoint with tuned open-source models and fine-tunes (Llama, DeepSeek, Qwen, etc.).

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: grok

const grok: (config) => AdapterFactory

Defined in: adapters/src/grok.ts:5

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: grokEmbedder

const grokEmbedder: (config) => EmbedFn

Defined in: adapters/src/embedders/grok.ts:5

#Parameters

#config

OpenAICompatibleEmbedderConfig

#Returns

EmbedFn

---

#Variable: groqAdapter

const groqAdapter: (config) => AdapterFactory = groq

Defined in: adapters/src/groq.ts:33

Alias for naming consistency with other native adapters.

Groq — OpenAI-compatible endpoint serving Llama / Mixtral on LPUs. Known for very low first-token latency.

Default model: openai/gpt-oss-120b.

#Parameters

#config

Partial<GroqConfig> & object

#Returns

AdapterFactory

---

#Variable: huggingface

const huggingface: (config) => AdapterFactory

Defined in: adapters/src/huggingface.ts:11

Hugging Face Inference Providers router. OpenAI-compatible endpoint that fans out to HF-hosted model inference providers. Pass any supported repo id as model, e.g. meta-llama/Meta-Llama-3.1-8B-Instruct.

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: kimi

const kimi: (config) => AdapterFactory

Defined in: adapters/src/kimi.ts:5

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: kimiEmbedder

const kimiEmbedder: (config) => EmbedFn

Defined in: adapters/src/embedders/kimi.ts:5

#Parameters

#config

OpenAICompatibleEmbedderConfig

#Returns

EmbedFn

---

#Variable: llamacpp

const llamacpp: (config) => AdapterFactory

Defined in: adapters/src/llamacpp.ts:9

llama.cpp's OpenAI-compatible HTTP server (llama-server). Defaults to http://localhost:8080/v1, matching the binary's default port.

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: lmstudio

const lmstudio: (config) => AdapterFactory

Defined in: adapters/src/lmstudio.ts:10

LM Studio local server. Start the built-in OpenAI-compatible server in LM Studio, then point this adapter at http://localhost:1234/v1 (the default).

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: mistral

const mistral: (config) => AdapterFactory

Defined in: adapters/src/mistral.ts:10

Mistral AI. Uses the OpenAI-compatible chat completions endpoint. Default models: mistral-large-latest, mistral-small-latest, open-mistral-nemo.

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: openrouter

const openrouter: (config) => AdapterFactory

Defined in: adapters/src/openrouter.ts:10

OpenRouter — routes to 300+ provider models behind a single OpenAI-compatible endpoint. Pass the fully-qualified id as model, e.g. anthropic/claude-sonnet-4-6.

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: qwen

const qwen: (config) => AdapterFactory = bail

Defined in: adapters/src/bail.ts:37

Alias matching Alibaba's product naming.

Alibaba Bailian (Qwen) via the DashScope OpenAI-compatibility endpoint.

Supports the Qwen-2.5 / 3 chat series and Qwen-VL multimodal models. APAC users typically prefer this over OpenAI for latency + data residency.

Default model: qwen-max.

#Parameters

#config

Partial<BailConfig> & object

#Returns

AdapterFactory

---

#Variable: replicateAdapter

const replicateAdapter: (config) => AdapterFactory = replicate

Defined in: adapters/src/replicate.ts:158

#Parameters

#config

ReplicateConfig

#Returns

AdapterFactory

---

#Variable: together

const together: (config) => AdapterFactory

Defined in: adapters/src/together.ts:9

Together AI. OpenAI-compatible endpoint, broad open-model catalog (Llama, Qwen, DeepSeek, Mixtral, etc.).

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: vertexAdapter

const vertexAdapter: (config) => AdapterFactory = vertex

Defined in: adapters/src/vertex.ts:76

#Parameters

#config

VertexConfig

#Returns

AdapterFactory

---

#Variable: vllm

const vllm: (config) => AdapterFactory

Defined in: adapters/src/vllm.ts:9

vLLM's OpenAI-compatible serving endpoint. Defaults to http://localhost:8000/v1 — the standard vllm serve ... port.

#Parameters

#config

OpenAICompatibleConfig

#Returns

AdapterFactory

---

#Variable: webllmAdapter

const webllmAdapter: (config) => AdapterFactory = webllm

Defined in: adapters/src/webllm.ts:144

#Parameters

#config

WebLlmConfig

#Returns

AdapterFactory

Explore nearby

• Peer
  @agentskit/adapters — Functions

  API functions for @agentskit/adapters.

• Peer
  @agentskit/adapters — Interfaces

  API interfaces for @agentskit/adapters.

• Peer
  @agentskit/adapters — Type Aliases

  API type aliases for @agentskit/adapters.