Documentation 2026-10-20 23:01 UTC to 2026-10-21 23:02 UTC

41| grok-4.20-0309-non-reasoning | T0: 37, T1: 50, T2: 75, T3: 125, T4: 208 | T0: 10M, T1: 15M, T2: 25M, T3: 45M, T4: 85M |41| grok-4.20-0309-non-reasoning | T0: 37, T1: 50, T2: 75, T3: 125, T4: 208 | T0: 10M, T1: 15M, T2: 25M, T3: 45M, T4: 85M |

42| grok-build-0.1 | T0: 37, T1: 50, T2: 75, T3: 125, T4: 208 | T0: 10M, T1: 15M, T2: 25M, T3: 45M, T4: 85M |42| grok-build-0.1 | T0: 37, T1: 50, T2: 75, T3: 125, T4: 208 | T0: 10M, T1: 15M, T2: 25M, T3: 45M, T4: 85M |

43| grok-4.20-multi-agent-0309 | T0: 9, T1: 12, T2: 18, T3: 31, T4: 56 | T0: 2.5M, T1: 3.7M, T2: 6.2M, T3: 11M, T4: 21M |43| grok-4.20-multi-agent-0309 | T0: 9, T1: 12, T2: 18, T3: 31, T4: 56 | T0: 2.5M, T1: 3.7M, T2: 6.2M, T3: 11M, T4: 21M |

44| grok-imagine-image-quality | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |

45| grok-imagine-image | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |44| grok-imagine-image | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |

46| grok-imagine-image-2.0 | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |45| grok-imagine-image-2.0 | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |

47| grok-imagine-video-1.5 | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |46| grok-imagine-image-quality | T0: 6, T1: 12, T2: 25, T3: 50, T4: 100 | — |

48| grok-imagine-video-1.5-lite | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |

49| grok-imagine-video | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |47| grok-imagine-video | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |

48| grok-imagine-video-1.5-lite | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |

49| grok-imagine-video-1.5 | T0: 10, T1: 20, T2: 39, T3: 79, T4: 158 | — |

50 50 

51**Voice & Audio**51**Voice & Audio**

52 52 

17* `frequency_penalty` (number | null) — (Not supported by reasoning models) Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.17* `frequency_penalty` (number | null) — (Not supported by reasoning models) Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.

18 18 

19* `guidelines` (array | null) — Per-category instructions for model output. See \[\`Guideline\`].19* `guidelines` (array | null) — Per-category instructions for model output. See \[\`Guideline\`].

20 Omitted requests leave this unset.20 Omitted requests leave this unset. A request whose guidelines cannot be

21 applied, for example on a multi-agent model or with \`stop\`, fails with

22 400\.

21 23 

22* `logit_bias` (object | null) — (Unsupported) A JSON object that maps tokens (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token.24* `logit_bias` (object | null) — (Unsupported) A JSON object that maps tokens (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token.

23 25 

78 80 

79 * `include_usage` (boolean, required) — Set an additional chunk to be streamed before the \`data: \[DONE]\` message. The other chunks will return \`null\` in \`usage\` field.81 * `include_usage` (boolean, required) — Set an additional chunk to be streamed before the \`data: \[DONE]\` message. The other chunks will return \`null\` in \`usage\` field.

80 82 

81* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.83* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Values above 1 are sampled at 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.

82 84 

83* `tool_choice` (string | object)85* `tool_choice` (string | object)

84 86 

86 88 

87* `top_logprobs` (integer | null) — An integer between 0 and 8 specifying the number of most likely tokens to return at each token position, each with an associated log probability. logprobs must be set to true if this parameter is used. Not supported by models \`grok-4.20\` and newer; the field will be silently ignored if set.89* `top_logprobs` (integer | null) — An integer between 0 and 8 specifying the number of most likely tokens to return at each token position, each with an associated log probability. logprobs must be set to true if this parameter is used. Not supported by models \`grok-4.20\` and newer; the field will be silently ignored if set.

88 90 

89* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both.91* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both. Values above the model's default top\_p, 0.95 for Grok models, are sampled at that default.

90 92 

91* `user` (string | null) — A unique identifier representing your end-user, which can help xAI to monitor and detect abuse.93* `user` (string | null) — A unique identifier representing your end-user, which can help xAI to monitor and detect abuse.

92 94 

201 201 

202* `system` (string | array\<object>)202* `system` (string | array\<object>)

203 203 

204* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. It may not work well with reasoning models.204* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Values above 1 are sampled at 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. It may not work well with reasoning models.

205 205 

206* `tool_choice` (object | object | object)206* `tool_choice` (object | object | object)

207 207 

209 209 

210* `top_k` (integer | null) — (Unsupported) When generating next tokens, randomly selecting the next token from the k most likely options.210* `top_k` (integer | null) — (Unsupported) When generating next tokens, randomly selecting the next token from the k most likely options.

211 211 

212* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both.212* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both. Values above the model's default top\_p, 0.95 for Grok models, are sampled at that default.

213 213 

214### Response Body214### Response Body

215 215 

17* `context_management` (array | null) — Optional context-management directives (e.g. compaction). Parsed but not yet executed.17* `context_management` (array | null) — Optional context-management directives (e.g. compaction). Parsed but not yet executed.

18 18 

19* `guidelines` (array | null) — Per-category instructions for model output. See \[\`Guideline\`].19* `guidelines` (array | null) — Per-category instructions for model output. See \[\`Guideline\`].

20 Omitted requests leave this unset.20 Omitted requests leave this unset. A request whose guidelines cannot be

21 applied, for example on a multi-agent model or with \`stop\`, fails with

22 400\.

21 23 

22* `include` (array | null) — What additional output data to include in the response. Supported values include24* `include` (array | null) — What additional output data to include in the response. Supported values include

23 \`reasoning.encrypted\_content\` (encrypted reasoning tokens) and tool-output options.25 \`reasoning.encrypted\_content\` (encrypted reasoning tokens) and tool-output options.

86 88 

87* `stream` (boolean | null) — If set, partial message deltas will be sent. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a \`data: \[DONE]\` message.89* `stream` (boolean | null) — If set, partial message deltas will be sent. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a \`data: \[DONE]\` message.

88 90 

89* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.91* `temperature` (number | null) — What sampling temperature to use, between 0 and 2. Values above 1 are sampled at 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.

90 92 

91* `text` (object)93* `text` (object)

92 94 

100 102 

101* `top_logprobs` (integer | null) — An integer between 0 and 8 specifying the number of most likely tokens to return at each token position, each with an associated log probability. logprobs must be set to true if this parameter is used. Not supported by models \`grok-4.20\` and newer; the field will be silently ignored if set.103* `top_logprobs` (integer | null) — An integer between 0 and 8 specifying the number of most likely tokens to return at each token position, each with an associated log probability. logprobs must be set to true if this parameter is used. Not supported by models \`grok-4.20\` and newer; the field will be silently ignored if set.

102 104 

103* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both.105* `top_p` (number | null) — An alternative to sampling with \`temperature\`, called nucleus sampling, where the model considers the results of the tokens with \`top\_p\` probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. It is generally recommended to alter this or \`temperature\` but not both. Values above the model's default top\_p, 0.95 for Grok models, are sampled at that default.

104 106 

105* `truncation` (string | null) — Not supported. Only maintained for compatibility reasons.107* `truncation` (string | null) — Not supported. Only maintained for compatibility reasons.

106 108