---
title: "Update Model"
method: PATCH
path: "/v1/accounts/{account_id}/models/{model_id}"
tags: ["Gateway"]
---

# Update Model

`PATCH /v1/accounts/{account_id}/models/{model_id}`

## Path parameters

- `account_id` string, required
- `model_id` string, required

## Request body

- object
  - `displayName` string — Human-readable display name of the model. e.g. "My Model" Must be fewer than 64 characters long.
  - `description` string — The description of the model. Must be fewer than 1000 characters long.
  - `createTime` string, date-time — The creation time of the model.
  - `state` 'STATE_UNSPECIFIED' | 'UPLOADING' | 'READY' — - UPLOADING: The model is still being uploaded (upload is asynchronous). - READY: The model is ready to be used.
  - `status` GatewayStatus
    - `code` 'OK' | 'CANCELLED' | 'UNKNOWN' | 'INVALID_ARGUMENT' | 'DEADLINE_EXCEEDED' | 'NOT_FOUND' | 'ALREADY_EXISTS' | 'PERMISSION_DENIED' | 'UNAUTHENTICATED' | 'RESOURCE_EXHAUSTED' | 'FAILED_PRECONDITION' | 'ABORTED' | 'OUT_OF_RANGE' | 'UNIMPLEMENTED' | 'INTERNAL' | 'UNAVAILABLE' | 'DATA_LOSS' — - OK: Not an error; returned on success. HTTP Mapping: 200 OK - CANCELLED: The operation was cancelled, typically by the caller. HTTP Mapping: 499 Client Closed Request - UNKNOWN: Unknown error. For example, this error may be returned when a `Status` value received from another address space belongs to an error space that is not known in this address space. Also errors raised by APIs that do not return enough error information may be converted to this error. HTTP Mapping: 500 Internal Server Error - INVALID_ARGUMENT: The client specified an invalid argument. Note that this differs from `FAILED_PRECONDITION`. `INVALID_ARGUMENT` indicates arguments that are problematic regardless of the state of the system (e.g., a malformed file name). HTTP Mapping: 400 Bad Request - DEADLINE_EXCEEDED: The deadline expired before the operation could complete. For operations that change the state of the system, this error may be returned even if the operation has completed successfully. For example, a successful response from a server could have been delayed long enough for the deadline to expire. HTTP Mapping: 504 Gateway Timeout - NOT_FOUND: Some requested entity (e.g., file or directory) was not found. Note to server developers: if a request is denied for an entire class of users, such as gradual feature rollout or undocumented allowlist, `NOT_FOUND` may be used. If a request is denied for some users within a class of users, such as user-based access control, `PERMISSION_DENIED` must be used. HTTP Mapping: 404 Not Found - ALREADY_EXISTS: The entity that a client attempted to create (e.g., file or directory) already exists. HTTP Mapping: 409 Conflict - PERMISSION_DENIED: The caller does not have permission to execute the specified operation. `PERMISSION_DENIED` must not be used for rejections caused by exhausting some resource (use `RESOURCE_EXHAUSTED` instead for those errors). `PERMISSION_DENIED` must not be used if the caller can not be identified (use `UNAUTHENTICATED` instead for those errors). This error code does not imply the request is valid or the requested entity exists or satisfies other pre-conditions. HTTP Mapping: 403 Forbidden - UNAUTHENTICATED: The request does not have valid authentication credentials for the operation. HTTP Mapping: 401 Unauthorized - RESOURCE_EXHAUSTED: Some resource has been exhausted, perhaps a per-user quota, or perhaps the entire file system is out of space. HTTP Mapping: 429 Too Many Requests - FAILED_PRECONDITION: The operation was rejected because the system is not in a state required for the operation's execution. For example, the directory to be deleted is non-empty, an rmdir operation is applied to a non-directory, etc. Service implementors can use the following guidelines to decide between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`: (a) Use `UNAVAILABLE` if the client can retry just the failing call. (b) Use `ABORTED` if the client should retry at a higher level. For example, when a client-specified test-and-set fails, indicating the client should restart a read-modify-write sequence. (c) Use `FAILED_PRECONDITION` if the client should not retry until the system state has been explicitly fixed. For example, if an "rmdir" fails because the directory is non-empty, `FAILED_PRECONDITION` should be returned since the client should not retry unless the files are deleted from the directory. HTTP Mapping: 400 Bad Request - ABORTED: The operation was aborted, typically due to a concurrency issue such as a sequencer check failure or transaction abort. See the guidelines above for deciding between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`. HTTP Mapping: 409 Conflict - OUT_OF_RANGE: The operation was attempted past the valid range. E.g., seeking or reading past end-of-file. Unlike `INVALID_ARGUMENT`, this error indicates a problem that may be fixed if the system state changes. For example, a 32-bit file system will generate `INVALID_ARGUMENT` if asked to read at an offset that is not in the range [0,2^32-1], but it will generate `OUT_OF_RANGE` if asked to read from an offset past the current file size. There is a fair bit of overlap between `FAILED_PRECONDITION` and `OUT_OF_RANGE`. We recommend using `OUT_OF_RANGE` (the more specific error) when it applies so that callers who are iterating through a space can easily look for an `OUT_OF_RANGE` error to detect when they are done. HTTP Mapping: 400 Bad Request - UNIMPLEMENTED: The operation is not implemented or is not supported/enabled in this service. HTTP Mapping: 501 Not Implemented - INTERNAL: Internal errors. This means that some invariants expected by the underlying system have been broken. This error code is reserved for serious errors. HTTP Mapping: 500 Internal Server Error - UNAVAILABLE: The service is currently unavailable. This is most likely a transient condition, which can be corrected by retrying with a backoff. Note that it is not always safe to retry non-idempotent operations. See the guidelines above for deciding between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`. HTTP Mapping: 503 Service Unavailable - DATA_LOSS: Unrecoverable data loss or corruption. HTTP Mapping: 500 Internal Server Error
    - `message` string — A developer-facing error message in English.
  - `kind` 'KIND_UNSPECIFIED' | 'HF_BASE_MODEL' | 'HF_PEFT_ADDON' | 'HF_TEFT_ADDON' | 'FLUMINA_BASE_MODEL' | 'FLUMINA_ADDON' | 'DRAFT_ADDON' | 'LIVE_MERGE' | 'CUSTOM_MODEL' | 'EMBEDDING_MODEL' | 'SNAPSHOT_MODEL' — - HF_BASE_MODEL: An LLM base model. - HF_PEFT_ADDON: A parameter-efficent fine-tuned addon. - HF_TEFT_ADDON: A token-eficient fine-tuned addon. - FLUMINA_BASE_MODEL: A Flumina base model. - FLUMINA_ADDON: A Flumina addon. - DRAFT_ADDON: A draft model used for speculative decoding in a deployment. - LIVE_MERGE: A live-merge model. - CUSTOM_MODEL: A customized model - EMBEDDING_MODEL: An Embedding model. - SNAPSHOT_MODEL: A snapshot model.
  - `githubUrl` string — The URL to GitHub repository of the model.
  - `huggingFaceUrl` string — The URL to the Hugging Face model.
  - `baseModelDetails` GatewayBaseModelDetails
    - `worldSize` integer — The default number of GPUs the model is served with. If not specified, the default is 1.
    - `checkpointFormat` 'CHECKPOINT_FORMAT_UNSPECIFIED' | 'NATIVE' | 'HUGGINGFACE' | 'UNINITIALIZED'
    - `huggingfaceFiles` string[] — A list of Hugging Face files associated with this model. Specified if and only if the checkpoint_format is HUGGINGFACE.
    - `parameterCount` string, int64 — The number of model parameters. For serverless models, this determines the price per token.
    - `moe` boolean — If true, this is a Mixture of Experts (MoE) model. For serverless models, this affects the price per token.
    - `tunable` boolean — Deprecated: V1 training stack only. Use per-category tunable flags on Model instead.
    - `modelType` string — The type of the model.
    - `supportsFireattention` boolean — Whether this model supports fireattention.
    - `defaultPrecision` 'PRECISION_UNSPECIFIED' | 'FP16' | 'FP8' | 'FP8_MM' | 'FP8_AR' | 'FP8_MM_KV_ATTN' | 'FP8_KV' | 'FP8_MM_V2' | 'FP8_V2' | 'FP8_MM_KV_ATTN_V2' | 'NF4' | 'FP4' | 'BF16' | 'FP4_BLOCKSCALED_MM' | 'FP4_MX_MOE'
    - `supportsMtp` boolean — If true, this model supports MTP.
  - `peftDetails` GatewayPEFTDetails
    - `baseModel` string, required
    - `r` integer, required — The rank of the update matrices. Must be between 4 and 64, inclusive.
    - `targetModules` string[], required
    - `baseModelType` string — The type of the model.
    - `mergeAddonModelName` string
  - `teftDetails` GatewayTEFTDetails
  - `public` boolean — If true, the model will be publicly readable.
  - `conversationConfig` GatewayConversationConfig
    - `style` string, required — The chat template to use.
    - `system` string — The system prompt (if the chat style supports it).
    - `template` string — The Jinja template (if style is "jinja").
  - `contextLength` integer — The maximum context length supported by the model.
  - `supportsImageInput` boolean — If set, images can be provided as input to the model.
  - `supportsTools` boolean — If set, tools (i.e. functions) can be provided as input to the model, and the model may respond with one or more tool calls.
  - `importedFrom` string — The name of the the model from which this was imported. This field is empty if the model was not imported.
  - `fineTuningJob` string — If the model was created from a fine-tuning job, this is the fine-tuning job name.
  - `defaultDraftModel` string — The default draft model to use when creating a deployment. If empty, speculative decoding is disabled by default.
  - `defaultDraftTokenCount` integer — The default draft token count to use when creating a deployment. Must be specified if default_draft_model is specified.
  - `deployedModelRefs` GatewayDeployedModelRef[] — Populated from GetModel API call only.
    - `name` string
    - `deployment` string — The resource name of the base deployment the model is deployed to.
    - `state` 'STATE_UNSPECIFIED' | 'UNDEPLOYING' | 'DEPLOYING' | 'DEPLOYED' | 'UPDATING' — - UNDEPLOYING: The model is being undeployed. - DEPLOYING: The model is being deployed. - DEPLOYED: The model is deployed and ready for inference. - UPDATING: there are updates happening with the deployed model
    - `default` boolean — If true, this is the default target when querying this model without the `#<deployment>` suffix. The first deployment a model is deployed to will have this field set to true automatically.
    - `public` boolean — If true, the deployed model will be publicly reachable.
  - `cluster` string — The resource name of the BYOC cluster to which this model belongs. e.g. accounts/my-account/clusters/my-cluster. Empty if it belongs to a Fireworks cluster.
  - `deprecationDate` TypeDate — * A full date, with non-zero year, month, and day values * A month and day value, with a zero year, such as an anniversary * A year on its own, with zero month and day values * A year and month value, with a zero day, such as a credit card expiration date Related types are [google.type.TimeOfDay][google.type.TimeOfDay] and `google.protobuf.Timestamp`.
    - `year` integer — Year of the date. Must be from 1 to 9999, or 0 to specify a date without a year.
    - `month` integer — Month of a year. Must be from 1 to 12, or 0 to specify a year without a month and day.
    - `day` integer — Day of a month. Must be from 1 to 31 and valid for the year and month, or 0 to specify a year by itself or a year and month where the day isn't significant.
  - `calibrated` boolean — If true, the model is calibrated and can be deployed to non-FP16 precisions.
  - `tunable` boolean — Deprecated: V1 training stack only — LoRA only, limited architecture support. If the model has use_training_v2=true, use supervised_lora_tunable and supervised_full_parameter_tunable instead.
  - `supportsLora` boolean — Whether this model supports LoRA.
  - `useHfApplyChatTemplate` boolean — If true, the model will use the Hugging Face apply_chat_template API to apply the chat template.
  - `updateTime` string, date-time — The update time for the model.
  - `defaultSamplingParams` object — A json object that contains the default sampling parameters for the model.
  - `rlTunable` boolean — Deprecated: V1 training stack only — LoRA only, limited architecture support. If the model has use_training_v2=true, use rl_lora_tunable and rl_full_parameter_tunable instead.
  - `trainingContextLength` integer — The maximum context length supported by the model.
  - `snapshotType` 'FULL_SNAPSHOT' | 'INCREMENTAL_SNAPSHOT'
  - `supportsServerless` boolean — If true, the model has a serverless deployment.
  - `useTrainingV2` boolean — If true, SFT jobs for this base model use service-mode (StatefulSet + orchestration sidecar) instead of the legacy batch Job path.
  - `supervisedLoraTunable` boolean — V2 only. Whether the model supports LoRA supervised fine-tuning and DPO (lora_rank > 0). True when a validated LORA_TRAINER training shape exists.
  - `supervisedFullParameterTunable` boolean — V2 only. Whether the model supports full-parameter supervised fine-tuning and DPO (lora_rank = 0). True when a validated POLICY_TRAINER training shape exists.
  - `rlLoraTunable` boolean — V2 only. Whether the model supports LoRA reinforcement learning (lora_rank > 0). True when a validated LORA_TRAINER training shape exists plus a deployment shape.
  - `rlFullParameterTunable` boolean — V2 only. Whether the model supports full-parameter reinforcement learning (lora_rank = 0). True when validated POLICY_TRAINER + LORA_TRAINER training shapes exist plus a deployment shape.
  - `encryptionState` 'ENCRYPTION_STATE_UNSPECIFIED' | 'ENCRYPTION_STATE_PLAINTEXT' | 'ENCRYPTION_STATE_CMEK' — EncryptionState is the authoritative, per-resource CMEK marker: the single source of truth for whether a resource's customer-data artifacts are encrypted. It is stamped at resource creation from the owning account's CMEK config and is immutable for the resource's life, so a resource always reads back with the state it was created under regardless of the account's current flag. The encrypt-on-write / decrypt-on-read decision for the resource's data keys off this field. - ENCRYPTION_STATE_UNSPECIFIED: Unstamped/legacy rows that predate CMEK. Treated as PLAINTEXT. - ENCRYPTION_STATE_PLAINTEXT: Customer-data artifacts are stored as plaintext (Fireworks-owned, no CMEK). - ENCRYPTION_STATE_CMEK: Customer-data artifacts are CMEK-encrypted; reads MUST decrypt (fail-closed).
  - `serverlessModes` GatewayServerlessMode[] — The serverless modes this base model is available on — the serverless read shape. Populated on read (e.g. the serverless catalog read path); one entry per serving mode. Empty for non-serverless models.
    - `name` string — The resource name, e.g. accounts/fireworks/models/kimi-k2p6/serverlessModes/fast. {AccountId} must be a serverless account (fireworks, ...); {ModelId} is the base model's id; {ServerlessModeId} is the mode name (default | priority | fast | spot).
    - `usageIdentifier` string — Invocation recipe: the value to pass as "model" in inference requests. Empty means use the base model id. May reference a DIFFERENT resource than the base model — a model OR a router, e.g. accounts/fireworks/routers/kimi-k2p6-fast for fast.
    - `serviceTier` string — Invocation recipe: the service_tier request-body flag layered on top of the model id (e.g. "priority"). Only the priority mode sets this today.
    - `skuInfos` GatewaySKUInfo[] — Per-path token pricing, sourced live from Orb on the read path. Reuses the same SKUInfo type as Model.sku_infos (e.g. "LLM input tokens (uncached)", "LLM input tokens (cached)", "LLM output tokens"; google.type.Money at unit="1M tokens"). Resolved per tier (DEFAULT/PRIORITY/FAST/...). Empty on the stored config; only populated by ListServerlessModels. (api_only) keeps this derived field out of the database entirely (like Model.sku_infos): it is never persisted by the admin write path and never read back from storage, only computed live at read time.
      - `sku` string
      - `amount` TypeMoney — Represents an amount of money with its currency type.
        - `currencyCode` string — The three-letter currency code defined in ISO 4217.
        - `units` string, int64 — The whole units of the amount. For example if `currencyCode` is `"USD"`, then 1 unit is one US dollar.
        - `nanos` integer — Number of nano (10^-9) units of the amount. The value must be between -999,999,999 and +999,999,999 inclusive. If `units` is positive, `nanos` must be positive or zero. If `units` is zero, `nanos` can be positive, zero, or negative. If `units` is negative, `nanos` must be negative or zero. For example $-1.75 is represented as `units`=-1 and `nanos`=-750,000,000.
      - `unit` string
    - `useCases` string[] — Curated use-case tags driving discovery filters (e.g. "coding").
    - `updatedBy` string — The user who last mutated this mode (audit).
    - `createTime` string, date-time
    - `updateTime` string, date-time

## Response `200`

A successful response.

- GatewayModel
  - `name` string
  - `displayName` string — Human-readable display name of the model. e.g. "My Model" Must be fewer than 64 characters long.
  - `description` string — The description of the model. Must be fewer than 1000 characters long.
  - `createTime` string, date-time — The creation time of the model.
  - `state` 'STATE_UNSPECIFIED' | 'UPLOADING' | 'READY' — - UPLOADING: The model is still being uploaded (upload is asynchronous). - READY: The model is ready to be used.
  - `status` GatewayStatus
    - `code` 'OK' | 'CANCELLED' | 'UNKNOWN' | 'INVALID_ARGUMENT' | 'DEADLINE_EXCEEDED' | 'NOT_FOUND' | 'ALREADY_EXISTS' | 'PERMISSION_DENIED' | 'UNAUTHENTICATED' | 'RESOURCE_EXHAUSTED' | 'FAILED_PRECONDITION' | 'ABORTED' | 'OUT_OF_RANGE' | 'UNIMPLEMENTED' | 'INTERNAL' | 'UNAVAILABLE' | 'DATA_LOSS' — - OK: Not an error; returned on success. HTTP Mapping: 200 OK - CANCELLED: The operation was cancelled, typically by the caller. HTTP Mapping: 499 Client Closed Request - UNKNOWN: Unknown error. For example, this error may be returned when a `Status` value received from another address space belongs to an error space that is not known in this address space. Also errors raised by APIs that do not return enough error information may be converted to this error. HTTP Mapping: 500 Internal Server Error - INVALID_ARGUMENT: The client specified an invalid argument. Note that this differs from `FAILED_PRECONDITION`. `INVALID_ARGUMENT` indicates arguments that are problematic regardless of the state of the system (e.g., a malformed file name). HTTP Mapping: 400 Bad Request - DEADLINE_EXCEEDED: The deadline expired before the operation could complete. For operations that change the state of the system, this error may be returned even if the operation has completed successfully. For example, a successful response from a server could have been delayed long enough for the deadline to expire. HTTP Mapping: 504 Gateway Timeout - NOT_FOUND: Some requested entity (e.g., file or directory) was not found. Note to server developers: if a request is denied for an entire class of users, such as gradual feature rollout or undocumented allowlist, `NOT_FOUND` may be used. If a request is denied for some users within a class of users, such as user-based access control, `PERMISSION_DENIED` must be used. HTTP Mapping: 404 Not Found - ALREADY_EXISTS: The entity that a client attempted to create (e.g., file or directory) already exists. HTTP Mapping: 409 Conflict - PERMISSION_DENIED: The caller does not have permission to execute the specified operation. `PERMISSION_DENIED` must not be used for rejections caused by exhausting some resource (use `RESOURCE_EXHAUSTED` instead for those errors). `PERMISSION_DENIED` must not be used if the caller can not be identified (use `UNAUTHENTICATED` instead for those errors). This error code does not imply the request is valid or the requested entity exists or satisfies other pre-conditions. HTTP Mapping: 403 Forbidden - UNAUTHENTICATED: The request does not have valid authentication credentials for the operation. HTTP Mapping: 401 Unauthorized - RESOURCE_EXHAUSTED: Some resource has been exhausted, perhaps a per-user quota, or perhaps the entire file system is out of space. HTTP Mapping: 429 Too Many Requests - FAILED_PRECONDITION: The operation was rejected because the system is not in a state required for the operation's execution. For example, the directory to be deleted is non-empty, an rmdir operation is applied to a non-directory, etc. Service implementors can use the following guidelines to decide between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`: (a) Use `UNAVAILABLE` if the client can retry just the failing call. (b) Use `ABORTED` if the client should retry at a higher level. For example, when a client-specified test-and-set fails, indicating the client should restart a read-modify-write sequence. (c) Use `FAILED_PRECONDITION` if the client should not retry until the system state has been explicitly fixed. For example, if an "rmdir" fails because the directory is non-empty, `FAILED_PRECONDITION` should be returned since the client should not retry unless the files are deleted from the directory. HTTP Mapping: 400 Bad Request - ABORTED: The operation was aborted, typically due to a concurrency issue such as a sequencer check failure or transaction abort. See the guidelines above for deciding between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`. HTTP Mapping: 409 Conflict - OUT_OF_RANGE: The operation was attempted past the valid range. E.g., seeking or reading past end-of-file. Unlike `INVALID_ARGUMENT`, this error indicates a problem that may be fixed if the system state changes. For example, a 32-bit file system will generate `INVALID_ARGUMENT` if asked to read at an offset that is not in the range [0,2^32-1], but it will generate `OUT_OF_RANGE` if asked to read from an offset past the current file size. There is a fair bit of overlap between `FAILED_PRECONDITION` and `OUT_OF_RANGE`. We recommend using `OUT_OF_RANGE` (the more specific error) when it applies so that callers who are iterating through a space can easily look for an `OUT_OF_RANGE` error to detect when they are done. HTTP Mapping: 400 Bad Request - UNIMPLEMENTED: The operation is not implemented or is not supported/enabled in this service. HTTP Mapping: 501 Not Implemented - INTERNAL: Internal errors. This means that some invariants expected by the underlying system have been broken. This error code is reserved for serious errors. HTTP Mapping: 500 Internal Server Error - UNAVAILABLE: The service is currently unavailable. This is most likely a transient condition, which can be corrected by retrying with a backoff. Note that it is not always safe to retry non-idempotent operations. See the guidelines above for deciding between `FAILED_PRECONDITION`, `ABORTED`, and `UNAVAILABLE`. HTTP Mapping: 503 Service Unavailable - DATA_LOSS: Unrecoverable data loss or corruption. HTTP Mapping: 500 Internal Server Error
    - `message` string — A developer-facing error message in English.
  - `kind` 'KIND_UNSPECIFIED' | 'HF_BASE_MODEL' | 'HF_PEFT_ADDON' | 'HF_TEFT_ADDON' | 'FLUMINA_BASE_MODEL' | 'FLUMINA_ADDON' | 'DRAFT_ADDON' | 'LIVE_MERGE' | 'CUSTOM_MODEL' | 'EMBEDDING_MODEL' | 'SNAPSHOT_MODEL' — - HF_BASE_MODEL: An LLM base model. - HF_PEFT_ADDON: A parameter-efficent fine-tuned addon. - HF_TEFT_ADDON: A token-eficient fine-tuned addon. - FLUMINA_BASE_MODEL: A Flumina base model. - FLUMINA_ADDON: A Flumina addon. - DRAFT_ADDON: A draft model used for speculative decoding in a deployment. - LIVE_MERGE: A live-merge model. - CUSTOM_MODEL: A customized model - EMBEDDING_MODEL: An Embedding model. - SNAPSHOT_MODEL: A snapshot model.
  - `githubUrl` string — The URL to GitHub repository of the model.
  - `huggingFaceUrl` string — The URL to the Hugging Face model.
  - `baseModelDetails` GatewayBaseModelDetails
    - `worldSize` integer — The default number of GPUs the model is served with. If not specified, the default is 1.
    - `checkpointFormat` 'CHECKPOINT_FORMAT_UNSPECIFIED' | 'NATIVE' | 'HUGGINGFACE' | 'UNINITIALIZED'
    - `huggingfaceFiles` string[] — A list of Hugging Face files associated with this model. Specified if and only if the checkpoint_format is HUGGINGFACE.
    - `parameterCount` string, int64 — The number of model parameters. For serverless models, this determines the price per token.
    - `moe` boolean — If true, this is a Mixture of Experts (MoE) model. For serverless models, this affects the price per token.
    - `tunable` boolean — Deprecated: V1 training stack only. Use per-category tunable flags on Model instead.
    - `modelType` string — The type of the model.
    - `supportsFireattention` boolean — Whether this model supports fireattention.
    - `defaultPrecision` 'PRECISION_UNSPECIFIED' | 'FP16' | 'FP8' | 'FP8_MM' | 'FP8_AR' | 'FP8_MM_KV_ATTN' | 'FP8_KV' | 'FP8_MM_V2' | 'FP8_V2' | 'FP8_MM_KV_ATTN_V2' | 'NF4' | 'FP4' | 'BF16' | 'FP4_BLOCKSCALED_MM' | 'FP4_MX_MOE'
    - `supportsMtp` boolean — If true, this model supports MTP.
  - `peftDetails` GatewayPEFTDetails
    - `baseModel` string, required
    - `r` integer, required — The rank of the update matrices. Must be between 4 and 64, inclusive.
    - `targetModules` string[], required
    - `baseModelType` string — The type of the model.
    - `mergeAddonModelName` string
  - `teftDetails` GatewayTEFTDetails
  - `public` boolean — If true, the model will be publicly readable.
  - `conversationConfig` GatewayConversationConfig
    - `style` string, required — The chat template to use.
    - `system` string — The system prompt (if the chat style supports it).
    - `template` string — The Jinja template (if style is "jinja").
  - `contextLength` integer — The maximum context length supported by the model.
  - `supportsImageInput` boolean — If set, images can be provided as input to the model.
  - `supportsTools` boolean — If set, tools (i.e. functions) can be provided as input to the model, and the model may respond with one or more tool calls.
  - `importedFrom` string — The name of the the model from which this was imported. This field is empty if the model was not imported.
  - `fineTuningJob` string — If the model was created from a fine-tuning job, this is the fine-tuning job name.
  - `defaultDraftModel` string — The default draft model to use when creating a deployment. If empty, speculative decoding is disabled by default.
  - `defaultDraftTokenCount` integer — The default draft token count to use when creating a deployment. Must be specified if default_draft_model is specified.
  - `deployedModelRefs` GatewayDeployedModelRef[] — Populated from GetModel API call only.
    - `name` string
    - `deployment` string — The resource name of the base deployment the model is deployed to.
    - `state` 'STATE_UNSPECIFIED' | 'UNDEPLOYING' | 'DEPLOYING' | 'DEPLOYED' | 'UPDATING' — - UNDEPLOYING: The model is being undeployed. - DEPLOYING: The model is being deployed. - DEPLOYED: The model is deployed and ready for inference. - UPDATING: there are updates happening with the deployed model
    - `default` boolean — If true, this is the default target when querying this model without the `#<deployment>` suffix. The first deployment a model is deployed to will have this field set to true automatically.
    - `public` boolean — If true, the deployed model will be publicly reachable.
  - `cluster` string — The resource name of the BYOC cluster to which this model belongs. e.g. accounts/my-account/clusters/my-cluster. Empty if it belongs to a Fireworks cluster.
  - `deprecationDate` TypeDate — * A full date, with non-zero year, month, and day values * A month and day value, with a zero year, such as an anniversary * A year on its own, with zero month and day values * A year and month value, with a zero day, such as a credit card expiration date Related types are [google.type.TimeOfDay][google.type.TimeOfDay] and `google.protobuf.Timestamp`.
    - `year` integer — Year of the date. Must be from 1 to 9999, or 0 to specify a date without a year.
    - `month` integer — Month of a year. Must be from 1 to 12, or 0 to specify a year without a month and day.
    - `day` integer — Day of a month. Must be from 1 to 31 and valid for the year and month, or 0 to specify a year by itself or a year and month where the day isn't significant.
  - `calibrated` boolean — If true, the model is calibrated and can be deployed to non-FP16 precisions.
  - `tunable` boolean — Deprecated: V1 training stack only — LoRA only, limited architecture support. If the model has use_training_v2=true, use supervised_lora_tunable and supervised_full_parameter_tunable instead.
  - `supportsLora` boolean — Whether this model supports LoRA.
  - `useHfApplyChatTemplate` boolean — If true, the model will use the Hugging Face apply_chat_template API to apply the chat template.
  - `updateTime` string, date-time — The update time for the model.
  - `defaultSamplingParams` object — A json object that contains the default sampling parameters for the model.
  - `rlTunable` boolean — Deprecated: V1 training stack only — LoRA only, limited architecture support. If the model has use_training_v2=true, use rl_lora_tunable and rl_full_parameter_tunable instead.
  - `trainingContextLength` integer — The maximum context length supported by the model.
  - `snapshotType` 'FULL_SNAPSHOT' | 'INCREMENTAL_SNAPSHOT'
  - `supportsServerless` boolean — If true, the model has a serverless deployment.
  - `useTrainingV2` boolean — If true, SFT jobs for this base model use service-mode (StatefulSet + orchestration sidecar) instead of the legacy batch Job path.
  - `supervisedLoraTunable` boolean — V2 only. Whether the model supports LoRA supervised fine-tuning and DPO (lora_rank > 0). True when a validated LORA_TRAINER training shape exists.
  - `supervisedFullParameterTunable` boolean — V2 only. Whether the model supports full-parameter supervised fine-tuning and DPO (lora_rank = 0). True when a validated POLICY_TRAINER training shape exists.
  - `rlLoraTunable` boolean — V2 only. Whether the model supports LoRA reinforcement learning (lora_rank > 0). True when a validated LORA_TRAINER training shape exists plus a deployment shape.
  - `rlFullParameterTunable` boolean — V2 only. Whether the model supports full-parameter reinforcement learning (lora_rank = 0). True when validated POLICY_TRAINER + LORA_TRAINER training shapes exist plus a deployment shape.
  - `encryptionState` 'ENCRYPTION_STATE_UNSPECIFIED' | 'ENCRYPTION_STATE_PLAINTEXT' | 'ENCRYPTION_STATE_CMEK' — EncryptionState is the authoritative, per-resource CMEK marker: the single source of truth for whether a resource's customer-data artifacts are encrypted. It is stamped at resource creation from the owning account's CMEK config and is immutable for the resource's life, so a resource always reads back with the state it was created under regardless of the account's current flag. The encrypt-on-write / decrypt-on-read decision for the resource's data keys off this field. - ENCRYPTION_STATE_UNSPECIFIED: Unstamped/legacy rows that predate CMEK. Treated as PLAINTEXT. - ENCRYPTION_STATE_PLAINTEXT: Customer-data artifacts are stored as plaintext (Fireworks-owned, no CMEK). - ENCRYPTION_STATE_CMEK: Customer-data artifacts are CMEK-encrypted; reads MUST decrypt (fail-closed).
  - `serverlessModes` GatewayServerlessMode[] — The serverless modes this base model is available on — the serverless read shape. Populated on read (e.g. the serverless catalog read path); one entry per serving mode. Empty for non-serverless models.
    - `name` string — The resource name, e.g. accounts/fireworks/models/kimi-k2p6/serverlessModes/fast. {AccountId} must be a serverless account (fireworks, ...); {ModelId} is the base model's id; {ServerlessModeId} is the mode name (default | priority | fast | spot).
    - `usageIdentifier` string — Invocation recipe: the value to pass as "model" in inference requests. Empty means use the base model id. May reference a DIFFERENT resource than the base model — a model OR a router, e.g. accounts/fireworks/routers/kimi-k2p6-fast for fast.
    - `serviceTier` string — Invocation recipe: the service_tier request-body flag layered on top of the model id (e.g. "priority"). Only the priority mode sets this today.
    - `skuInfos` GatewaySKUInfo[] — Per-path token pricing, sourced live from Orb on the read path. Reuses the same SKUInfo type as Model.sku_infos (e.g. "LLM input tokens (uncached)", "LLM input tokens (cached)", "LLM output tokens"; google.type.Money at unit="1M tokens"). Resolved per tier (DEFAULT/PRIORITY/FAST/...). Empty on the stored config; only populated by ListServerlessModels. (api_only) keeps this derived field out of the database entirely (like Model.sku_infos): it is never persisted by the admin write path and never read back from storage, only computed live at read time.
      - `sku` string
      - `amount` TypeMoney — Represents an amount of money with its currency type.
        - `currencyCode` string — The three-letter currency code defined in ISO 4217.
        - `units` string, int64 — The whole units of the amount. For example if `currencyCode` is `"USD"`, then 1 unit is one US dollar.
        - `nanos` integer — Number of nano (10^-9) units of the amount. The value must be between -999,999,999 and +999,999,999 inclusive. If `units` is positive, `nanos` must be positive or zero. If `units` is zero, `nanos` can be positive, zero, or negative. If `units` is negative, `nanos` must be negative or zero. For example $-1.75 is represented as `units`=-1 and `nanos`=-750,000,000.
      - `unit` string
    - `useCases` string[] — Curated use-case tags driving discovery filters (e.g. "coding").
    - `updatedBy` string — The user who last mutated this mode (audit).
    - `createTime` string, date-time
    - `updateTime` string, date-time

---

[API](https://skmtc.net/fireworks/apis/fireworks-ai-anthropic-compatible-messages-api.md) · [All operations](https://skmtc.net/fireworks/apis/fireworks-ai-anthropic-compatible-messages-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/fireworks/fireworks-ai-anthropic-compatible-messages-api/versions/954d6bc5d922/schema)