Important: This documentation covers Yarn 1 (Classic).
For Yarn 2+ docs and migration guide, see yarnpkg.com.

Package detail

@ai-sdk/openai-compatible

vercel1.8mApache-2.01.0.22TypeScript support: included

This package provides a foundation for implementing providers that expose an OpenAI-compatible API.

ai

readme

packages/ai/README.md

changelog

@ai-sdk/openai-compatible

1.0.22

Patch Changes

  • f02b7ab: Add textVerbosity provider option support

1.0.21

Patch Changes

1.0.20

Patch Changes

1.0.19

Patch Changes

1.0.18

Patch Changes

  • 28363da: feat(openai-compatible): add supportsStructuredOutputs to provider settings

1.0.17

Patch Changes

  • 3aed04c: feat(provider/openai-compatible): set user-agent header for createOpenAICompatible

1.0.16

Patch Changes

1.0.15

Patch Changes

1.0.14

Patch Changes

  • 818f021: Prevent redundant reasoningEffort field in request body (in favor of reasoning_effort)

1.0.13

Patch Changes

1.0.12

Patch Changes

1.0.11

Patch Changes

1.0.10

Patch Changes

  • 7ca3aee: refactoring(provider/openai-compatible): move models into folders

1.0.9

Patch Changes

1.0.8

Patch Changes

  • 515c891: fix duplicated tool-input-start in certain cases

1.0.7

Patch Changes

1.0.6

Patch Changes

1.0.5

Patch Changes

  • 8f8a521: fix(providers): use convertToBase64 for Uint8Array image parts to produce valid data URLs; keep mediaType normalization and URL passthrough
  • e92b78b: fix (provider/openai-compatible): revert usage fallback specific to moonshotai

1.0.4

Patch Changes

  • 5f4c71f: feat (provider/openai-compatible): fall back to look for usage in choices
  • da314cd: chore (provider/openai-compatible): inline usage fallback logic

1.0.3

Patch Changes

  • a0934f8: feat (provider/openai-compatible): look for reasoning in 'reasoning' field as well

1.0.2

Patch Changes

1.0.1

Patch Changes

  • 0e8ed8e: feat(provider/openai-compatible): Update OpenAI compat embedding schema to support providerMetadata

1.0.0

Major Changes

  • d5f588f: AI SDK 5

  • 516be5b: ### Move Image Model Settings into generate options

    Image Models no longer have settings. Instead, maxImagesPerCall can be passed directly to generateImage(). All other image settings can be passed to providerOptions[provider].

    Before

    await generateImage({
  model: luma.image('photon-flash-1', {
    maxImagesPerCall: 5,
    pollIntervalMillis: 500,
  }),
  prompt,
  n: 10,
});

    After

    await generateImage({
  model: luma.image('photon-flash-1'),
  prompt,
  n: 10,
  maxImagesPerCall: 5,
  providerOptions: {
    luma: { pollIntervalMillis: 5 },
  },
});

    Pull Request: https://github.com/vercel/ai/pull/6180

Patch Changes

  • 6db02c9: chore(openai-compatible): remove simulateStreaming
  • fa49207: feat(providers/openai-compatible): convert to providerOptions
  • cf8280e: fix(providers/xai): return actual usage when streaming instead of NaN
  • b9a6121: fix (provider/openai-compatible): change tool_call type schema to nullish
  • e2aceaf: feat: add raw chunk support
  • db72adc: chore(providers/openai): update completion model to use providerOptions
  • 26735b5: chore(embedding-model): add v2 interface
  • 443d8ec: feat(embedding-model-v2): add response body field
  • 42e32b0: feat(providers/xai): add reasoningEffort provider option
  • 7b069ed: allow any string as reasoningEffort
  • 66962ed: fix(packages): export node10 compatible types
  • d9209ca: fix (image-model): specificationVersion: v1 -> v2
  • 9301f86: refactor (image-model): rename ImageModelV1 to ImageModelV2
  • 737f1e2: Add (optional) includeUsage option to createOpenAICompatible
  • d1a034f: feature: using Zod 4 for internal stuff
  • fd65bc6: chore(embedding-model-v2): rename rawResponse to response
  • 1b101e1: feat (provider/openai-compatible): allow providers that do not specify function type
  • 205077b: fix: improve Zod compatibility
  • 281bb1c: fix (provider/openai-incompatible): support empty tool calls
  • Updated dependencies [a571d6e]
  • Updated dependencies [742b7be]
  • Updated dependencies [e7fcc86]
  • Updated dependencies [7cddb72]
  • Updated dependencies [ccce59b]
  • Updated dependencies [e2b9e4b]
  • Updated dependencies [95857aa]
  • Updated dependencies [45c1ea2]
  • Updated dependencies [6f6bb89]
  • Updated dependencies [060370c]
  • Updated dependencies [dc714f3]
  • Updated dependencies [b5da06a]
  • Updated dependencies [d1a1aa1]
  • Updated dependencies [63f9e9b]
  • Updated dependencies [5d142ab]
  • Updated dependencies [d5f588f]
  • Updated dependencies [e025824]
  • Updated dependencies [0571b98]
  • Updated dependencies [b6b43c7]
  • Updated dependencies [4fef487]
  • Updated dependencies [48d257a]
  • Updated dependencies [0c0c0b3]
  • Updated dependencies [0d2c085]
  • Updated dependencies [40acf9b]
  • Updated dependencies [9222aeb]
  • Updated dependencies [e2aceaf]
  • Updated dependencies [411e483]
  • Updated dependencies [8ba77a7]
  • Updated dependencies [7b3ae3f]
  • Updated dependencies [a166433]
  • Updated dependencies [26735b5]
  • Updated dependencies [443d8ec]
  • Updated dependencies [a8c8bd5]
  • Updated dependencies [abf9a79]
  • Updated dependencies [14c9410]
  • Updated dependencies [e86be6f]
  • Updated dependencies [9bf7291]
  • Updated dependencies [2e13791]
  • Updated dependencies [9f95b35]
  • Updated dependencies [66962ed]
  • Updated dependencies [0d06df6]
  • Updated dependencies [472524a]
  • Updated dependencies [dd3ff01]
  • Updated dependencies [d9c98f4]
  • Updated dependencies [05d2819]
  • Updated dependencies [9301f86]
  • Updated dependencies [0a87932]
  • Updated dependencies [c4a2fec]
  • Updated dependencies [957b739]
  • Updated dependencies [79457bd]
  • Updated dependencies [a3f768e]
  • Updated dependencies [7435eb5]
  • Updated dependencies [8aa9e20]
  • Updated dependencies [4617fab]
  • Updated dependencies [ac34802]
  • Updated dependencies [0054544]
  • Updated dependencies [cb68df0]
  • Updated dependencies [ad80501]
  • Updated dependencies [68ecf2f]
  • Updated dependencies [9e9c809]
  • Updated dependencies [32831c6]
  • Updated dependencies [6dc848c]
  • Updated dependencies [6b98118]
  • Updated dependencies [d0f9495]
  • Updated dependencies [63d791d]
  • Updated dependencies [87b828f]
  • Updated dependencies [3f2f00c]
  • Updated dependencies [bfdca8d]
  • Updated dependencies [0ff02bb]
  • Updated dependencies [7979f7f]
  • Updated dependencies [39a4fab]
  • Updated dependencies [44f4aba]
  • Updated dependencies [9bd5ab5]
  • Updated dependencies [57edfcb]
  • Updated dependencies [faf8446]
  • Updated dependencies [7ea4132]
  • Updated dependencies [d1a034f]
  • Updated dependencies [5c56081]
  • Updated dependencies [fd65bc6]
  • Updated dependencies [023ba40]
  • Updated dependencies [ea7a7c9]
  • Updated dependencies [26535e0]
  • Updated dependencies [e030615]
  • Updated dependencies [5e57fae]
  • Updated dependencies [393138b]
  • Updated dependencies [c57e248]
  • Updated dependencies [88a8ee5]
  • Updated dependencies [41fa418]
  • Updated dependencies [205077b]
  • Updated dependencies [71f938d]
  • Updated dependencies [3795467]
  • Updated dependencies [28a5ed5]
  • Updated dependencies [7182d14]
  • Updated dependencies [c1e6647]
  • Updated dependencies [1766ede]
  • Updated dependencies [811dff3]
  • Updated dependencies [f10304b]
  • Updated dependencies [dd5fd43]
  • Updated dependencies [33f4a6a]
  • Updated dependencies [383cbfa]
  • Updated dependencies [27deb4d]
  • Updated dependencies [c4df419]

1.0.0-beta.13

Patch Changes

1.0.0-beta.12

Patch Changes

1.0.0-beta.11

Patch Changes

1.0.0-beta.10

Patch Changes

1.0.0-beta.9

Patch Changes

1.0.0-beta.8

Patch Changes

1.0.0-beta.7

Patch Changes

1.0.0-beta.6

Patch Changes

  • 281bb1c: fix (provider/openai-incompatible): support empty tool calls

1.0.0-beta.5

Patch Changes

1.0.0-beta.4

Patch Changes

  • 1b101e1: feat (provider/openai-compatible): allow providers that do not specify function type

1.0.0-beta.3

Patch Changes

  • 7b069ed: allow any string as reasoningEffort

1.0.0-beta.2

Patch Changes

  • d1a034f: feature: using Zod 4 for internal stuff
  • Updated dependencies [0571b98]
  • Updated dependencies [39a4fab]
  • Updated dependencies [d1a034f]

1.0.0-beta.1

Patch Changes

  • Updated dependencies [742b7be]
  • Updated dependencies [7cddb72]
  • Updated dependencies [ccce59b]
  • Updated dependencies [e2b9e4b]
  • Updated dependencies [45c1ea2]
  • Updated dependencies [e025824]
  • Updated dependencies [0d06df6]
  • Updated dependencies [472524a]
  • Updated dependencies [dd3ff01]
  • Updated dependencies [7435eb5]
  • Updated dependencies [cb68df0]
  • Updated dependencies [bfdca8d]
  • Updated dependencies [44f4aba]
  • Updated dependencies [023ba40]
  • Updated dependencies [5e57fae]
  • Updated dependencies [71f938d]
  • Updated dependencies [28a5ed5]

1.0.0-alpha.15

Patch Changes

1.0.0-alpha.14

Patch Changes

1.0.0-alpha.13

Patch Changes

1.0.0-alpha.12

Patch Changes

1.0.0-alpha.11

Patch Changes

1.0.0-alpha.10

Patch Changes

1.0.0-alpha.9

Patch Changes

1.0.0-alpha.8

Patch Changes

1.0.0-alpha.7

Patch Changes

1.0.0-alpha.6

Patch Changes

1.0.0-alpha.4

Patch Changes

1.0.0-alpha.3

Patch Changes

1.0.0-alpha.2

Patch Changes

1.0.0-alpha.1

Patch Changes

1.0.0-canary.19

Patch Changes

1.0.0-canary.18

Patch Changes

1.0.0-canary.17

Major Changes

  • 516be5b: ### Move Image Model Settings into generate options

    Image Models no longer have settings. Instead, maxImagesPerCall can be passed directly to generateImage(). All other image settings can be passed to providerOptions[provider].

    Before

    await generateImage({
  model: luma.image('photon-flash-1', {
    maxImagesPerCall: 5,
    pollIntervalMillis: 500,
  }),
  prompt,
  n: 10,
});

    After

    await generateImage({
  model: luma.image('photon-flash-1'),
  prompt,
  n: 10,
  maxImagesPerCall: 5,
  providerOptions: {
    luma: { pollIntervalMillis: 5 },
  },
});

    Pull Request: https://github.com/vercel/ai/pull/6180

Patch Changes

1.0.0-canary.16

Patch Changes

1.0.0-canary.15

Patch Changes

1.0.0-canary.14

Patch Changes

1.0.0-canary.13

Patch Changes

1.0.0-canary.12

Patch Changes

1.0.0-canary.11

Patch Changes

  • db72adc: chore(providers/openai): update completion model to use providerOptions
  • 42e32b0: feat(providers/xai): add reasoningEffort provider option
  • 66962ed: fix(packages): export node10 compatible types
  • 9301f86: refactor (image-model): rename ImageModelV1 to ImageModelV2
  • Updated dependencies [66962ed]
  • Updated dependencies [9301f86]
  • Updated dependencies [a3f768e]

1.0.0-canary.10

Patch Changes

1.0.0-canary.9

Patch Changes

1.0.0-canary.8

Patch Changes

1.0.0-canary.7

Patch Changes

  • fa49207: feat(providers/openai-compatible): convert to providerOptions
  • 26735b5: chore(embedding-model): add v2 interface
  • 443d8ec: feat(embedding-model-v2): add response body field
  • fd65bc6: chore(embedding-model-v2): rename rawResponse to response
  • Updated dependencies [26735b5]
  • Updated dependencies [443d8ec]
  • Updated dependencies [14c9410]
  • Updated dependencies [d9c98f4]
  • Updated dependencies [c4a2fec]
  • Updated dependencies [0054544]
  • Updated dependencies [9e9c809]
  • Updated dependencies [32831c6]
  • Updated dependencies [d0f9495]
  • Updated dependencies [fd65bc6]
  • Updated dependencies [393138b]
  • Updated dependencies [7182d14]

1.0.0-canary.6

Patch Changes

1.0.0-canary.5

Patch Changes

1.0.0-canary.4

Patch Changes

1.0.0-canary.3

Patch Changes

1.0.0-canary.2

Patch Changes

1.0.0-canary.1

Patch Changes

1.0.0-canary.0

Major Changes

  • d5f588f: AI SDK 5

Patch Changes

0.2.5

Patch Changes

  • d186cca: feat (provider/openai-compatible): add additional token usage metrics

0.2.4

Patch Changes

0.2.3

Patch Changes

0.2.2

Patch Changes

  • a6b55cc: feat (providers/openai-compatible): add openai-compatible image model and use as xai image model base

0.2.1

Patch Changes

0.2.0

Minor Changes

  • 5bc638d: AI SDK 4.2

Patch Changes

0.1.17

Patch Changes

0.1.16

Patch Changes

0.1.15

Patch Changes

0.1.14

Patch Changes

0.1.13

Patch Changes

0.1.12

Patch Changes

0.1.11

Patch Changes

0.1.10

Patch Changes

0.1.9

Patch Changes

0.1.8

Patch Changes

0.1.7

Patch Changes

  • f2c6c37: feat (provider/openai-compatible): support providerOptions in generateText/streamText

0.1.6

Patch Changes

0.1.5

Patch Changes

0.1.4

Patch Changes

0.1.3

Patch Changes

  • 361fd08: chore: update a few add'l processor references to extractor

0.1.2

Patch Changes

0.1.1

Patch Changes

0.1.0

Minor Changes

  • 62ba5ad: release: AI SDK 4.1

Patch Changes

0.0.18

Patch Changes

0.0.17

Patch Changes

  • ae57beb: feat (provider/openai-compatible): add support for optional custom URL parameters in requests.

0.0.16

Patch Changes

  • 7611964: feat (provider/xai): Support structured output for latest models.

0.0.15

Patch Changes

0.0.14

Patch Changes

0.0.13

Patch Changes

  • 6564812: feat (provider/openai-compatible): Add'l exports for customization.

0.0.12

Patch Changes

  • 70003b8: feat (provider/openai-compatible): Allow extending messages via metadata.

0.0.11

Patch Changes

0.0.10

Patch Changes

  • baae8f4: feat (provider/deepinfra): Add DeepInfra provider.

0.0.9

Patch Changes

  • 9c7653b: feat (docs): Update OpenAI Compatible docs for new package.

0.0.8

Patch Changes

  • 6faab13: feat (provider/openai-compatible): simulated streaming setting

0.0.7

Patch Changes

  • ad2bf11: feat (provider/fireworks): Add Fireworks provider.

0.0.6

Patch Changes

0.0.5

Patch Changes

  • e958996: fix (provider/openai-compatible): remove unused index property from validation

0.0.4

Patch Changes

0.0.3

Patch Changes

  • a9a19cb: fix (provider/openai,groq): prevent sending duplicate tool calls

0.0.2

Patch Changes

  • fc18132: feat (ai/core): experimental output for generateText

0.0.1

Patch Changes

  • 962978b: feat (packages/openai-compatible): Base for OpenAI-compatible providers.