local-first · v0.1

DeepSeek

DeepSeek reasoning/chat models.

llmexperimental#775c1439

Supported adapters (this provider can fulfill)

One adapter supports many providers. The default is switchable without changing any workflow.

Served capabilities (derived via adapters)

embeddingllmreasoningspeechvision

Authentication & secrets

authenticationapi_key
auth note
required secretsDEEPSEEK_API_KEY
permissions

Secrets are referenced by name only — never values. This runtime performs no OAuth and stores no credentials.

Approval & compliance

requires approvalfalse
approval kinds
compliance
enterprise readyfalse

Pricing

modelper_token
unitUSD / 1K tokens
note

Latency & limits

latencymedium
max concurrency
rate limit / min
quotaunbounded (request)

Availability & regions

globaltrue
regionsglobal
healthunknown

Deployment targets (without redesign)

cloud
execution modesapi, cloud

Dependencies & future dispatch

depends on

Providers describe themselves only. Selection and execution are owned by the future Dispatch Runtime — this layer dispatches nothing.