Skip to content
HIPS
  • Earn with your GPU
  • Pricing
  • Docs
Sign inJoin early access
⌘K

Get started

  1. 01Developer hub
  2. 02Quickstart
  3. 03Migrate to HIPS

Guides

  1. 04HIPS SDK
  2. 05Import and run your model
  3. 06Model and task guides
    1. ATask guides
    2. BLanguage (chat)
    3. CClassification
    4. DImage
    5. EVideo
    6. FAudio
    7. GDocument processing
  4. 07Batch jobs
  5. 08Prompt caching
  6. 09Code with HIPS
  7. 10Coding-tool setup
    1. AHIPS Code
    2. BClaude Code
    3. CCodex
    4. DOpenCode
    5. EKilo
    6. FPi
    7. GQwen Code
    8. HCrush
    9. ICline
    10. JPrime Agent
    11. KAider
    12. LGoose
    13. MHermes Agent
    14. NGrok Build
    15. ODeepSeek Harness
    16. POther coding tools and editor integrations
  8. 11Command line
    1. AInstall and update
    2. BSign in
    3. CStatus and models
    4. DCoding agents
  9. 12Console
    1. AAccount
    2. BWorkspaces
    3. CAPI keys
    4. DSettings
    5. EUsage
    6. FModels
    7. GPlayground
    8. HDeployments
  10. 13Billing
  11. 14Provider setup
    1. AWhen it works
    2. BCommands
    3. CSettings
    4. DProvider Console
    5. EEarnings
    6. FCrash reports
    7. GData centers
  12. 15Semantic routing
  13. 16Guardrails
  14. 17Knowledge

Reference

  1. 18API reference
    1. AAuthentication
    2. BChat completions
    3. CResponses
    4. DMessages
    5. ECount tokens
    6. FModels
    7. GPrices API
    8. HTasks
    9. IEmbeddings and rerank
    10. JImages
    11. KJobs
    12. LUploads
    13. MFiles
    14. NBatches
    15. OWebhooks
    16. PVector stores
    17. QDeployments
    18. RRequirements and headers
    19. SErrors
    20. TUsage
    21. UVersioning
  2. 19Operating guidance
    1. ALimits
    2. BErrors
    3. CRetries and billing outcomes
    4. DIdempotency
    5. EStreaming
    6. FData handling
    7. GCompatibility matrix
  3. 20Changelog
Sheet indexAPI reference

Get started

  1. 01Developer hub
  2. 02Quickstart
  3. 03Migrate to HIPS

Guides

  1. 04HIPS SDK
  2. 05Import and run your model
  3. 06Model and task guides
    1. ATask guides
    2. BLanguage (chat)
    3. CClassification
    4. DImage
    5. EVideo
    6. FAudio
    7. GDocument processing
  4. 07Batch jobs
  5. 08Prompt caching
  6. 09Code with HIPS
  7. 10Coding-tool setup
    1. AHIPS Code
    2. BClaude Code
    3. CCodex
    4. DOpenCode
    5. EKilo
    6. FPi
    7. GQwen Code
    8. HCrush
    9. ICline
    10. JPrime Agent
    11. KAider
    12. LGoose
    13. MHermes Agent
    14. NGrok Build
    15. ODeepSeek Harness
    16. POther coding tools and editor integrations
  8. 11Command line
    1. AInstall and update
    2. BSign in
    3. CStatus and models
    4. DCoding agents
  9. 12Console
    1. AAccount
    2. BWorkspaces
    3. CAPI keys
    4. DSettings
    5. EUsage
    6. FModels
    7. GPlayground
    8. HDeployments
  10. 13Billing
  11. 14Provider setup
    1. AWhen it works
    2. BCommands
    3. CSettings
    4. DProvider Console
    5. EEarnings
    6. FCrash reports
    7. GData centers
  12. 15Semantic routing
  13. 16Guardrails
  14. 17Knowledge

Reference

  1. 18API reference
    1. AAuthentication
    2. BChat completions
    3. CResponses
    4. DMessages
    5. ECount tokens
    6. FModels
    7. GPrices API
    8. HTasks
    9. IEmbeddings and rerank
    10. JImages
    11. KJobs
    12. LUploads
    13. MFiles
    14. NBatches
    15. OWebhooks
    16. PVector stores
    17. QDeployments
    18. RRequirements and headers
    19. SErrors
    20. TUsage
    21. UVersioning
  2. 19Operating guidance
    1. ALimits
    2. BErrors
    3. CRetries and billing outcomes
    4. DIdempotency
    5. EStreaming
    6. FData handling
    7. GCompatibility matrix
  3. 20Changelog
18
  1. Docs/
  2. API reference/
  3. Chat completions
(Detail B)
llms.txt

Chat completions

On this sheet7 sections
  1. Three rules
  2. The request
  3. The model field
  4. The response
  5. Streaming
  6. Tool calls and structured outputs
  7. Errors this route can answer

For open-weight language models the public API is OpenAI-compatible, so OpenAI client SDKs work with a HIPS key, base URL and model id. Compatibility is a request and response shape, validated per endpoint, model and client version; it is never a model list.

POST/v1/chat/completionsserved

Chat completion in OpenAI's shape, streamed or not.

F
Fig. 1, sheet 18API reference

Three rules

  • Nothing is silently dropped. A field that would change the model's output or the bill and that HIPS or the named model cannot honor is a 400, never ignored.
  • Nothing is silently substituted. The model served is the model named. No fallback to another model, ever. A router name in model asks HIPS to choose the model, and the answer names the model chosen in model, x-hips-model and x-hips-router.
  • A requirement is never weakened to make a request succeed.

The request

The body is OpenAI's. A 400 names the field in param and the model in message when the refusal depends on the model.

Parts list: Fields

  1. 01
    modelRequired

    A catalog id, an alias, a pinned id or a dep_ id; a catalog id, alias or pinned id may end in :auto, :certified or :community to name its pool (A pool in the model id).

  2. 02
    messagesRequired, non-empty

    Roles system, developer, user, assistant, tool. content a string or an array of parts. Text parts always; image_url parts only on a model that advertises vision; input_audio and file parts are a 400 unsupported_parameter (audio goes through the task API).

  3. 03
    streamAccepted

    Server-sent events, one data: line per chunk, then data: [DONE].

  4. 04
    stream_optionsAccepted with stream: true

    include_usage honored. Without stream a 400 unsupported_value.

  5. 05
    max_completion_tokens, max_tokensAccepted

    Both names; both given and different is a 400 invalid_value. Over the model's max_output_tokens is a 400 max_tokens_exceeded. Absent: the model's default from its entry.

  6. 06
    temperature, top_p, stop, seed, presence_penalty, frequency_penaltyAccepted

    Passed to the engine. stop holds at most 16 entries (decided 2026-11-01); more is a 400 invalid_value. seed is best effort; determinism is not promised across workers.

  7. 07
    nAccepted as 1

    Any other value is a 400 unsupported_value: one answer per request, so the quote and the bill are one line.

  8. 08
    response_formatAccepted

    text always; json_object on a model advertising json_object; json_schema with name, schema and strict on a model advertising json_schema. Otherwise a 400 unsupported_parameter naming the feature.

  9. 09
    tools, tool_choiceAccepted

    On a model advertising tools. Shapes are OpenAI's, unchanged.

  10. 10
    parallel_tool_callsAccepted

    A permission, as OpenAI means it: true, OpenAI's default, lets the model make more than one tool call in a turn where it advertises parallel_tool_calls in the pool the request is placed in; where it does not, the request is served as if it said false (at most one call, the engine told false), never refused for it. false asks for at most one call and is accepted on every model with tools. A model without tools still refuses tools (since 2026-11-11).

  11. 11
    logprobs, top_logprobsAccepted

    On a model advertising logprobs.

  12. 12
    logit_biasAccepted

    On a model advertising logit_bias.

  13. 13
    reasoning_effortAccepted

    On a model advertising reasoning; otherwise a 400 unsupported_parameter.

  14. 14
    userAccepted, discarded

    It changes neither the output nor the bill. Not stored. Stated here so the discard is not silent.

  15. 15
    prompt_cache_keyAccepted

    A string of at most 512 characters (default proposed: 512, open under A-O6); longer is a 400 invalid_value. It travels to the worker with the body and is never logged or stored. It does nothing more yet: no cached token is billed today (Prompt caching).

  16. 16
    cache_saltRejected, 400 unknown_parameter

    The prompt cache's salt is the gateway's to derive, never the client's.

  17. 17
    prompt_cache_options, prompt_cache_breakpoint, prompt_cache_retentionRejected, 400 unsupported_parameter

    HIPS promises no cache lifetime and takes no breakpoint.

  18. 18
    store, metadataRejected, 400 stateful_not_supported

    The gateway stores no bodies.

  19. 19
    functions, function_callRejected, 400 unsupported_parameter

    OpenAI's deprecated names; tools is the shape.

  20. 20
    modalities, audio, prediction, web_search_options, service_tierRejected, 400 unsupported_parameter

    Audio and media go through the task API; pools and pricing modes are headers, not a tier. One exception: service_tier: "flex" on a :certified id runs the request in the Community pool, for any caller, and the answer echoes it (The combined feed).

  21. 21
    Any other nameRejected, 400 unknown_parameter

    Typos are caught, not ignored.

The model field

ValueFormResolves to
Catalog id<family>-<size>[-<variant>], lowercaseThe current pinned version of that catalog entry; moves with releases
AliasAs listed in the entry's hips.aliasesThe entry it names
Pinned id<catalog id>-<version>That version, as long as it is served
Id with a pool<catalog id>:auto, :certified or :community (an alias or pinned id the same way)That model in Auto, or in that pool alone with no fallback, as x-hips-pool names it (A pool in the model id)
Deployment iddep_<base32>A customer model deployment, or a catalog model pinned to an always-on allocation, of the key's workspace
  • Value
    Catalog id
    Form
    <family>-<size>[-<variant>], lowercase
    Resolves to
    The current pinned version of that catalog entry; moves with releases
  • Value
    Alias
    Form
    As listed in the entry's hips.aliases
    Resolves to
    The entry it names
  • Value
    Pinned id
    Form
    <catalog id>-<version>
    Resolves to
    That version, as long as it is served
  • Value
    Id with a pool
    Form
    <catalog id>:auto, :certified or :community (an alias or pinned id the same way)
    Resolves to
    That model in Auto, or in that pool alone with no fallback, as x-hips-pool names it (A pool in the model id)
  • Value
    Deployment id
    Form
    dep_<base32>
    Resolves to
    A customer model deployment, or a catalog model pinned to an always-on allocation, of the key's workspace
SituationAnswer
Unknown id, or another workspace's dep_404 model_not_found
A model or a router the workspace's plan does not include403 model_not_on_plan, with feature and required_plan (Errors)
A model an access guardrail of the workspace, the key or the person keeps out403 model_not_allowed, naming the guardrail (Guardrails)
A key whose kind excludes the route (a retrieval key)403 key_scope_denied
A dep_ whose model is not a language model400 not_a_language_model (use POST /v1/tasks)
A deployment in stopped400 deployment_stopped
A deployment in allocating or loading503 deployment_not_ready with Retry-After
A deployment waiting for its version's tariff (tariff_pending)503 deployment_not_ready with Retry-After; the message says the tariff is pending. Nothing is charged
A deployment in degradedServed; the state is in GET /v1/models/{id}
  • Situation
    Unknown id, or another workspace's dep_
    Answer
    404 model_not_found
  • Situation
    A model or a router the workspace's plan does not include
    Answer
    403 model_not_on_plan, with feature and required_plan (Errors)
  • Situation
    A model an access guardrail of the workspace, the key or the person keeps out
    Answer
    403 model_not_allowed, naming the guardrail (Guardrails)
  • Situation
    A key whose kind excludes the route (a retrieval key)
    Answer
    403 key_scope_denied
  • Situation
    A dep_ whose model is not a language model
    Answer
    400 not_a_language_model (use POST /v1/tasks)
  • Situation
    A deployment in stopped
    Answer
    400 deployment_stopped
  • Situation
    A deployment in allocating or loading
    Answer
    503 deployment_not_ready with Retry-After
  • Situation
    A deployment waiting for its version's tariff (tariff_pending)
    Answer
    503 deployment_not_ready with Retry-After; the message says the tariff is pending. Nothing is charged
  • Situation
    A deployment in degraded
    Answer
    Served; the state is in GET /v1/models/{id}

The response's model and the x-hips-model header carry the pinned id that was served; x-hips-deployment carries the dep_ id when the request named one.

The response

OpenAI's chat.completion object: id, object, created, model, choices[{index, message{role, content, tool_calls, refusal}, finish_reason, logprobs}], usage.

Parts list: HIPS specifics

  1. 01
    id

    The request id, req_..., so a log line and a support question name the same thing.

  2. 02
    model

    The pinned id served.

  3. 03
    finish_reason

    stop, length, tool_calls. content_filter is never emitted: HIPS applies no output filter on this route.

  4. 04
    system_fingerprint

    Omitted.

  5. 05
    usage

    The usage object. Money is never in it.

  6. 06
    Extras

    In headers, not in the body, except the streamed hips object.

Streaming

stream: true answers 200 with Content-Type: text/event-stream, one data: line per chunk in OpenAI's chat.completion.chunk shape, then data: [DONE]. The rules are on Streaming; the final usage chunk with its hips object looks like this:

Final chunk

00000000000000000000000000000000000000001data: {"id":"req_...","object":"chat.completion.chunk","created":0,"model":"<pinned id>","choices":[],"usage":{"prompt_tokens":0,"completion_tokens":0,"total_tokens":0},"hips":{"cost_micros":0,"latency_ms":0,"quote":"quote_..."}}2 3data: [DONE]
Final chunk sample

Tool calls and structured outputs

Features are advertised per model in GET /v1/models and come from the engine and the model together. Shapes are OpenAI's, unchanged. A feature the model does not advertise is a 400 unsupported_parameter.

FeatureRequestResponse
toolstools: [{"type":"function","function":{...}}]; tool_choice as auto, none, required or a named functionmessage.tool_calls with arguments a JSON string; finish_reason: "tool_calls". Streamed as delta.tool_calls by index
parallel_tool_callsparallel_tool_calls: true (or left out, as OpenAI's default); without the feature true is served as false, never refusedMore than one entry in tool_calls
json_objectresponse_format: {"type":"json_object"}content is one JSON document
json_schemaresponse_format: {"type":"json_schema","json_schema":{"name","schema","strict": true}}content validates against the schema; strict: false accepted where the model advertises it
visionimage_url parts in a user messageAs text
logprobslogprobs: true, top_logprobschoices[].logprobs
reasoningreasoning_effortusage.completion_tokens_details.reasoning_tokens
logit_biaslogit_biasApplied by the engine
  • Feature
    tools
    Request
    tools: [{"type":"function","function":{...}}]; tool_choice as auto, none, required or a named function
    Response
    message.tool_calls with arguments a JSON string; finish_reason: "tool_calls". Streamed as delta.tool_calls by index
  • Feature
    parallel_tool_calls
    Request
    parallel_tool_calls: true (or left out, as OpenAI's default); without the feature true is served as false, never refused
    Response
    More than one entry in tool_calls
  • Feature
    json_object
    Request
    response_format: {"type":"json_object"}
    Response
    content is one JSON document
  • Feature
    json_schema
    Request
    response_format: {"type":"json_schema","json_schema":{"name","schema","strict": true}}
    Response
    content validates against the schema; strict: false accepted where the model advertises it
  • Feature
    vision
    Request
    image_url parts in a user message
    Response
    As text
  • Feature
    logprobs
    Request
    logprobs: true, top_logprobs
    Response
    choices[].logprobs
  • Feature
    reasoning
    Request
    reasoning_effort
    Response
    usage.completion_tokens_details.reasoning_tokens
  • Feature
    logit_bias
    Request
    logit_bias
    Response
    Applied by the engine

Errors this route can answer

StatusTypeCodeSituationRetry
400invalid_request_errorinvalid_jsonThe body is not JSON, or not an objectNo
400invalid_request_errormissing_required_parameterA required field or header is absent; param names itNo
400invalid_request_errorinvalid_typeA field has the wrong JSON typeNo
400invalid_request_errorinvalid_valueA value is out of range or malformed: stop with more than 16 entries, a prompt_cache_key over 512 characters, a string holding a lone surrogate escape, a router name with an empty or unknown suffix, both max_tokens names with different valuesNo
400invalid_request_errorunknown_parameterA field the gateway does not know, at any depth; param names itNo
400invalid_request_errorunsupported_parameterA known field HIPS does not accept, or a feature the named model does not advertise in the pool the request resolves to; param names it and message names the modelNo
400invalid_request_errorunsupported_valuen other than 1, stream_options without stream, a message role, part, item or tool type not accepted, a hosted tool that is not servedNo
400invalid_request_errorstateful_not_supportedState HIPS does not keep: store and metadata on chat completions; store: true, previous_response_id, conversation, metadata, background: true and item_reference on ResponsesNo
400invalid_request_errorcontext_length_exceededThe prompt is over the model's context_length, or over the largest window any GPU serves it at now, or held only in a size that the GPUs able to hold it keep failing to read within the time limit; that limit and the prompt's count in the message, the limit the largest size that works now for this request, never more than the pool's context_window on Models; the prompt is never truncated. On embeddings and rerank, an input longer than the model reads at once, never cutNo
400invalid_request_errormax_tokens_exceededmax_tokens (on Responses max_output_tokens) over the model's max_output_tokensNo
400invalid_request_errorinvalid_imageAn image part or input cannot be decoded, is of an unsupported type or over the pixel limitNo
400invalid_request_errornot_a_language_modelA task model named on an OpenAI-compatible routeNo
400invalid_request_errordeployment_stoppedThe dep_ is in stopped; start it in the ConsoleNo
400invalid_request_errorinvalid_requirementA requirement header or field with an unknown valueNo
400invalid_request_errorrequirement_conflictTwo requirements that cannot both hold, a header and a hips field that differ, a pool the model is never offered in, community with an always-on deployment, a knowledge base whose requirements the request cannot meetNo
400invalid_request_errorstable_price_not_in_forcepricing_mode: stable with no stable price covering this workspace, model and periodNo
400invalid_request_errorrouter_no_candidateA router has no model it can choose for this request: none takes what the request carries (param the first field at fault) or none is allowed under its requirements (param the header); the message says why and what would serve it. Before any worker, unchargedNo
400invalid_request_errorguardrail_blockedA content moderation guardrail with block found prompt injection or a sensitive pattern; param names the message and the error's guardrail member names the guardrail, the check and the pattern's type, never the matched text. Before any model ran, unchargedNo
402insufficient_creditinsufficient_creditThe balance does not cover the request's maximum cost at the bound quoteAfter a top-up
402insufficient_creditbudget_exceededA budget guardrail of the workspace, the key or the person is at its limit with block; the error carries guardrail and resets_at (null for a budget that never resets). Never a payment challengeAfter resets_at, or once the limit is raised
403permission_errormodel_not_on_planA model or feature the workspace's plan does not include, read from the plan's terms: on Free a model outside the Free plan's models (today gemma-4-e4b-it-8b, gpt-oss-120b, gpt-oss-20b and qwen3-14b), a router or Knowledge; on Free and Pro a deployment (dep_). The error adds feature and required_plan, the lowest plan that includes it. Answered before any placement, uncharged. A deployment refused this way is kept and answers again once the workspace is on a plan with deploymentsAfter an owner or admin chooses a plan that includes it
403permission_errorpool_not_allowedA pool the workspace may not use for this model: one an access guardrail keeps out (named in the error), or a customer model's pool whose runtime-access conditions were not acceptedNo
403permission_errormodel_not_allowed, region_not_allowedA model, or a region requirement, that an access guardrail keeps out, the guardrail named in the error's guardrail memberNo
404not_found_errormodel_not_foundUnknown catalog id, alias, pinned id or dep_; another workspace's dep_; a name starting with hips-router that is not one of the four routersNo
429rate_limit_errorrate_limit_exceededA rate limit; param is requests, tokens, uploads, batches (batches created per hour) or knowledge_search; Retry-AfterYes, after Retry-After
503overloaded_errorno_capacityNo eligible worker under the request's requirements, in any pool the requirements allow; Retry-After. The message says what those pools can do now: a GPU that fits the model takes it when free or loads it when asked, which can take minutes or longer; or nothing can serve it, and a retry sooner than the given delay gets the same answerYes, after the given delay
503overloaded_errormodel_loadingEvery GPU that holds the model would have to load it first, and none could give a first token within the request's bound now, so the request is not sent to one; HIPS's placement brings the model up for the requests that follow; Retry-After. Nothing is chargedYes, after the given delay
503overloaded_errorcontext_size_not_measuredThe prompt fits a window a GPU can hold, in a size not measured on it yet (never tried there, or being measured); Retry-After, and the largest size that works now in the message, as context_length_exceeded names it. A size the GPUs keep failing to read in time is the 400 context_length_exceeded instead, since a retry would not succeed. A prompt's size is never answered no_capacityYes, or send a shorter prompt
503overloaded_errorpool_unavailableThe named pool has no eligible supply for this model now and fallback is off; the other pool may have; Retry-AfterYes, or with fallback
503overloaded_errordeployment_not_readyThe dep_ is allocating or loading, or waits for its version's tariff (tariff_pending); Retry-AfterYes
503overloaded_erroroverloadedEligible workers exist and every one is saturated, or the queue is full; Retry-AfterYes
503overloaded_errorguardrail_unavailableA content moderation check with block could not run in time, or is one a decision service judges on a request that requires a region (Guardrails); Retry-After; unchargedYes, unless the request requires a region
500api_errorinternal_errorA fault in HIPS. Nothing billedOnce
500api_errorattempt_failedEvery attempt failed for a reason that is not the client's and retries are exhausted. Nothing billedOnce
500api_errordeadline_exceededThe request's own time bound ended it; x-should-retry: false. Nothing billedNot unchanged
  • Status
    400
    Type
    invalid_request_error
    Code
    invalid_json
    Situation
    The body is not JSON, or not an object
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    missing_required_parameter
    Situation
    A required field or header is absent; param names it
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    invalid_type
    Situation
    A field has the wrong JSON type
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    invalid_value
    Situation
    A value is out of range or malformed: stop with more than 16 entries, a prompt_cache_key over 512 characters, a string holding a lone surrogate escape, a router name with an empty or unknown suffix, both max_tokens names with different values
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    unknown_parameter
    Situation
    A field the gateway does not know, at any depth; param names it
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    unsupported_parameter
    Situation
    A known field HIPS does not accept, or a feature the named model does not advertise in the pool the request resolves to; param names it and message names the model
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    unsupported_value
    Situation
    n other than 1, stream_options without stream, a message role, part, item or tool type not accepted, a hosted tool that is not served
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    stateful_not_supported
    Situation
    State HIPS does not keep: store and metadata on chat completions; store: true, previous_response_id, conversation, metadata, background: true and item_reference on Responses
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    context_length_exceeded
    Situation
    The prompt is over the model's context_length, or over the largest window any GPU serves it at now, or held only in a size that the GPUs able to hold it keep failing to read within the time limit; that limit and the prompt's count in the message, the limit the largest size that works now for this request, never more than the pool's context_window on Models; the prompt is never truncated. On embeddings and rerank, an input longer than the model reads at once, never cut
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    max_tokens_exceeded
    Situation
    max_tokens (on Responses max_output_tokens) over the model's max_output_tokens
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    invalid_image
    Situation
    An image part or input cannot be decoded, is of an unsupported type or over the pixel limit
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    not_a_language_model
    Situation
    A task model named on an OpenAI-compatible route
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    deployment_stopped
    Situation
    The dep_ is in stopped; start it in the Console
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    invalid_requirement
    Situation
    A requirement header or field with an unknown value
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    requirement_conflict
    Situation
    Two requirements that cannot both hold, a header and a hips field that differ, a pool the model is never offered in, community with an always-on deployment, a knowledge base whose requirements the request cannot meet
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    stable_price_not_in_force
    Situation
    pricing_mode: stable with no stable price covering this workspace, model and period
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    router_no_candidate
    Situation
    A router has no model it can choose for this request: none takes what the request carries (param the first field at fault) or none is allowed under its requirements (param the header); the message says why and what would serve it. Before any worker, uncharged
    Retry
    No
  • Status
    400
    Type
    invalid_request_error
    Code
    guardrail_blocked
    Situation
    A content moderation guardrail with block found prompt injection or a sensitive pattern; param names the message and the error's guardrail member names the guardrail, the check and the pattern's type, never the matched text. Before any model ran, uncharged
    Retry
    No
  • Status
    402
    Type
    insufficient_credit
    Code
    insufficient_credit
    Situation
    The balance does not cover the request's maximum cost at the bound quote
    Retry
    After a top-up
  • Status
    402
    Type
    insufficient_credit
    Code
    budget_exceeded
    Situation
    A budget guardrail of the workspace, the key or the person is at its limit with block; the error carries guardrail and resets_at (null for a budget that never resets). Never a payment challenge
    Retry
    After resets_at, or once the limit is raised
  • Status
    403
    Type
    permission_error
    Code
    model_not_on_plan
    Situation
    A model or feature the workspace's plan does not include, read from the plan's terms: on Free a model outside the Free plan's models (today gemma-4-e4b-it-8b, gpt-oss-120b, gpt-oss-20b and qwen3-14b), a router or Knowledge; on Free and Pro a deployment (dep_). The error adds feature and required_plan, the lowest plan that includes it. Answered before any placement, uncharged. A deployment refused this way is kept and answers again once the workspace is on a plan with deployments
    Retry
    After an owner or admin chooses a plan that includes it
  • Status
    403
    Type
    permission_error
    Code
    pool_not_allowed
    Situation
    A pool the workspace may not use for this model: one an access guardrail keeps out (named in the error), or a customer model's pool whose runtime-access conditions were not accepted
    Retry
    No
  • Status
    403
    Type
    permission_error
    Code
    model_not_allowed, region_not_allowed
    Situation
    A model, or a region requirement, that an access guardrail keeps out, the guardrail named in the error's guardrail member
    Retry
    No
  • Status
    404
    Type
    not_found_error
    Code
    model_not_found
    Situation
    Unknown catalog id, alias, pinned id or dep_; another workspace's dep_; a name starting with hips-router that is not one of the four routers
    Retry
    No
  • Status
    429
    Type
    rate_limit_error
    Code
    rate_limit_exceeded
    Situation
    A rate limit; param is requests, tokens, uploads, batches (batches created per hour) or knowledge_search; Retry-After
    Retry
    Yes, after Retry-After
  • Status
    503
    Type
    overloaded_error
    Code
    no_capacity
    Situation
    No eligible worker under the request's requirements, in any pool the requirements allow; Retry-After. The message says what those pools can do now: a GPU that fits the model takes it when free or loads it when asked, which can take minutes or longer; or nothing can serve it, and a retry sooner than the given delay gets the same answer
    Retry
    Yes, after the given delay
  • Status
    503
    Type
    overloaded_error
    Code
    model_loading
    Situation
    Every GPU that holds the model would have to load it first, and none could give a first token within the request's bound now, so the request is not sent to one; HIPS's placement brings the model up for the requests that follow; Retry-After. Nothing is charged
    Retry
    Yes, after the given delay
  • Status
    503
    Type
    overloaded_error
    Code
    context_size_not_measured
    Situation
    The prompt fits a window a GPU can hold, in a size not measured on it yet (never tried there, or being measured); Retry-After, and the largest size that works now in the message, as context_length_exceeded names it. A size the GPUs keep failing to read in time is the 400 context_length_exceeded instead, since a retry would not succeed. A prompt's size is never answered no_capacity
    Retry
    Yes, or send a shorter prompt
  • Status
    503
    Type
    overloaded_error
    Code
    pool_unavailable
    Situation
    The named pool has no eligible supply for this model now and fallback is off; the other pool may have; Retry-After
    Retry
    Yes, or with fallback
  • Status
    503
    Type
    overloaded_error
    Code
    deployment_not_ready
    Situation
    The dep_ is allocating or loading, or waits for its version's tariff (tariff_pending); Retry-After
    Retry
    Yes
  • Status
    503
    Type
    overloaded_error
    Code
    overloaded
    Situation
    Eligible workers exist and every one is saturated, or the queue is full; Retry-After
    Retry
    Yes
  • Status
    503
    Type
    overloaded_error
    Code
    guardrail_unavailable
    Situation
    A content moderation check with block could not run in time, or is one a decision service judges on a request that requires a region (Guardrails); Retry-After; uncharged
    Retry
    Yes, unless the request requires a region
  • Status
    500
    Type
    api_error
    Code
    internal_error
    Situation
    A fault in HIPS. Nothing billed
    Retry
    Once
  • Status
    500
    Type
    api_error
    Code
    attempt_failed
    Situation
    Every attempt failed for a reason that is not the client's and retries are exhausted. Nothing billed
    Retry
    Once
  • Status
    500
    Type
    api_error
    Code
    deadline_exceeded
    Situation
    The request's own time bound ended it; x-should-retry: false. Nothing billed
    Retry
    Not unchanged

Related sheets

  • Quickstart
  • Streaming
  • Compatibility matrix
Previous sheetAuthentication
Next sheetResponses
Sheet
18 of 20
Detail
B
Updated
2026-11-15
Sign in to the Console

On this sheet

  1. Three rules
  2. The request
  3. The model field
  4. The response
  5. Streaming
  6. Tool calls and structured outputs
  7. Errors this route can answer

Product

  • Inference
  • Models
  • Deploy your model
  • Always-on GPU capacity
  • How it works
  • Use cases
  • Pricing

Developer

  • Docs
  • Quickstart
  • Migrate to HIPS
  • HIPS SDK
  • Batch jobs
  • API reference

Provider

  • Earn with your GPU
  • Register GPU interest
  • Download for Mac
  • Download for Windows
  • Download for Linux

Resource

  • Security and Data
  • Resources
  • llms.txt

Compare

  • OpenRouter alternatives
  • Together AI alternatives
  • Groq alternatives
  • All comparisons

Company

  • About
  • Contact
  • Affiliate program
  • HIPS Payments

Legal

  • Legal
  • Consumer Health Data Privacy Policy
  • Report content
  • Cookie preferences

Support

  • Contact HIPS
  • Looking for payments support?

HIPS. AI inference for builders.

hips.com

Hyperscale Inference Processing Systems OÜ