Together AI RL API is one of 28 APIs that Together AI publishes on the APIs.io network, described by a machine-readable OpenAPI specification and an AsyncAPI event-driven specification.
Tagged areas include RL. The published artifact set on APIs.io includes an OpenAPI specification, API documentation, an API reference, and an AsyncAPI specification.
This API exposes
15 operations
across 14 paths,
and defines 50 schemas.
It is described by OpenAPI 3.2.0, at version 2.0.0.
Requests are made against a single base URL, https://api.together.ai/v1.
15 operations14 paths50 schemas8 GET7 POST
Metadata
The identity and technical contract details declared by the specification.
Together AI RL API declares
1 security scheme
for authenticating requests.
It accepts HTTP bearer tokens (bearerAuth).
By default, every request must be authenticated.
Paths & Operations 15
Across 14 paths, the API surfaces 15 operations — 8 GET, 7 POST. Each is listed below with its method, path, parameters, and response codes.
The contract defines 50 schemas that model the data the API accepts and returns. The most detailed are RL.TrainingSession (11 properties), RL.SamplingParams (6 properties), RL.ForwardBackwardOperation (4 properties), RL.TrainingCheckpointOperation (4 properties). Each schema is shown below with its type and property counts.
RL.InferenceCheckpointOperation
object
Async inference checkpoint operation
4 properties2 required
RL.InferenceCheckpointResult
object
Result of an inference checkpoint operation
1 property1 required
RL.TrainingOperationError
object
Error details for a failed training operation
2 properties
RL.LossType
string
RL.OptimStepBody
object
2 properties
RL.LossInputs
object
3 properties1 required
RL.SessionType
string
Type of a training session. TRAINERANDGENERATOR provisions both trainer and generator; TRAINERONLY provisions only the trainer and rejects generator-dependent…
RL.TrainingSample
object
2 properties2 required
RL.LoraConfig
object
LoRA adapter configuration
3 properties
RL.ListMeta
object
Pagination metadata for list responses
3 properties
RL.CheckpointDownloadResponse
object
Presigned download URLs for a checkpoint's files
1 property1 required
RL.TrainingSession
object
A training session and its current state
11 properties10 required
RL.InputChunk
object
1 property
RL.GRPOLossParams
object
4 properties
RL.GRPOLossInputs
object
3 properties2 required
RL.SampleOperation
object
Async sample operation
4 properties2 required
RL.ModelInput
object
1 property1 required
RL.OptimStepResult
object
Result of an optimizer step operation
1 property1 required
RL.TrainingCheckpointResult
object
Result of a save training checkpoint operation
1 property1 required
RL.ForwardBackwardResult
object
Result of a forward-backward pass operation
2 properties1 required
RL.LossTargetTokens
object
2 properties1 required
RL.SamplingParams
object
6 properties
RL.DType
string
RL.TrainingOperationErrorCode
string
RL.InferenceCheckpoint
object
Saved inference checkpoint
4 properties3 required
RL.ForwardBackwardOperation
object
Async forward-backward pass operation
4 properties2 required
RL.OptimStepOperation
object
Async optimizer step operation
4 properties2 required
RL.TrainingCheckpointOperation
object
Async save training checkpoint operation
4 properties2 required
RL.SampleSequence
object
A single generated completion sequence with tokens and logprobs
3 properties1 required
RL.InferenceCheckpointRegistration
object
Model registration details for an inference checkpoint
2 properties2 required
RL.LossConfig
object
3 properties1 required
RL.TrainingOperationStatus
string
RL.LossLogprobs
object
2 properties1 required
RL.ForwardBackwardBody
object
2 properties2 required
RL.SampleRollout
object
Completions generated for a single prompt
1 property1 required
RL.CheckpointFile
object
A downloadable file within a checkpoint
3 properties3 required
RL.SampleBody
object
3 properties1 required
RL.TrainingSessionStatus
string
Status of the training session
RL.StartTrainingSessionRequest
object
4 properties1 required
RL.LossAdvantages
object
2 properties1 required
RL.GRPOLossAggregationType
string
RL.CheckpointVariant
string
Checkpoint variant: merged (full model) or adapter (LoRA weights only)
ErrorData
object
1 property1 required
RL.AdamWOptimizerParams
object
AdamW optimizer parameters
4 properties
RL.TrainingCheckpoint
object
Saved training checkpoint
3 properties3 required
RL.EncodedText
object
1 property1 required
RL.SampleResult
object
Result of a sample operation
1 property1 required
RL.CrossEntropyLossParams
object
Cross-entropy loss parameters (currently empty).
RL.TrainingSessionsListResponse
object
Paginated list of training sessions
2 properties
RL.LossMask
object
Per-token loss mask (1=compute loss, 0=ignore)
2 properties1 required
Specification
The full machine-readable OpenAPI contract behind this narrative.
Every API here is available over the API and to AI agents over MCP. APIs is not yet its own endpoint on the v1 API. Reach this content through network search and the tag graph, or the MCP server below.
Installs https://mcp.apievangelist.com/mcp in Claude, Cursor, VS Code and the rest — one button, every client.
MCP tools for apis
4 tools reach this content
search_api_evangelistSearch every content type across the network at once.
find_relatedThe shared-tag relevance graph — what else covers this.
get_tagEverything one tag labels, across all content types.
guide_topicPRO — a curated bundle for a topic: area, guidance, rules, papers, stories, services.
A second provider on the same verified email joins the account you already have.
Your account
ⓘWhere this information came from
This is an independent, third-party profile of Together AI RL API, published by
API Evangelist. We do not operate, host, resell, or
support these APIs, and we are not affiliated with or endorsed by the company unless stated above.
Everything here is built from publicly available information — the company's own site,
developer portal, documentation, public repositories, and the specifications it publishes for public use.
Nothing is obtained by breaching a system, defeating an access control, or using credentials.
The Kin Score and Agent Readiness rating are independently calculated assessments of a company's
public API artifacts, scored against a published rubric. They are not certifications,
endorsements, security assessments, or audits.
Corrections, re-scores, and removal are free — no partnership or purchase required, and
you do not need to justify the request. A removed company is recorded as unrated, never scored
zero for having asked. Acknowledgement within one business day; removal within two.
info@apievangelist.com
·
Read the full data-sourcing policy → On a security or compliance team? Put security in the subject line and
you will get a person, not a form — we will tell you exactly which public URLs this profile was built from.
