Build with Cheaper Inference

Quick start

1. Create an account.
2. Add at least $5 to your wallet. On your first funding, we add $10 in bonus credit automatically.
3. Create an API key from the dashboard and copy it once.
4. Set the base URL for your SDK or app to the compatible endpoint below.
5. Put a supported model name in the model field.
6. Send your first request and review usage in the dashboard.

Chat completion

1curl https://api.cheaperinference.com/v1/chat/completions \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "claude-opus-4.6",6    "messages": [{"role": "user", "content": "Hello!"}]7  }'

1from openai import OpenAI23client = OpenAI(4    api_key="ci_live_YOUR_API_KEY",5    base_url="https://api.cheaperinference.com/v1",6)78response = client.chat.completions.create(9    model="claude-opus-4.6",10    messages=[{"role": "user", "content": "Hello!"}],11)

1import OpenAI from "openai";23const client = new OpenAI({4  apiKey: "ci_live_YOUR_API_KEY",5  baseURL: "https://api.cheaperinference.com/v1"6});78const response = await client.chat.completions.create({9  model: "gpt-5.4",10  messages: [{ role: "user", content: "Hello!" }]11});

cURL Responses request

1curl https://api.cheaperinference.com/v1/responses \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "gpt-5.4",6    "input": "Hello!",7    "store": false,8    "stream": true9  }'

The stateless Responses compatibility layer supports Codex streaming, function calls, and custom local tools such as apply_patch. Requests must use store: false. Stored responses, previous_response_id, conversations, background mode, and provider-hosted web or file search are not supported.

API reference

Every endpoint, header, and error code, with runnable request builders. The generated code updates as you edit the fields, so it is useful to read and copy whether or not you are signed in. The sections after it cover topics this reference does not, and go deeper on a few it only summarises.

Available models and pricing

Copy the exact model ID into the model field. Every model we serve, with its current price and capabilities, is on the live model catalog. Each GET /v1/models item also includes an aliases array with the vendor-qualified id, including vendor-specific version punctuation, so migrations do not need a hand-maintained map. IDs absent from this catalog return HTTP 404.

How pricing and routing work

For every request, Cheaper Inference ranks the eligible provider routes by estimated cost and starts with the lowest-cost route available for that model and request. If that provider is unavailable, has insufficient capacity, or fails before output begins, the gateway automatically tries the next-lowest-cost eligible route. Your public model ID and API response format remain unchanged.

A fallback route can have a different rate from the first route. The final customer charge follows the route that successfully serves the request, but it will never exceed the model maker's applicable direct API list price. Cheaper Inference adds no separate routing surcharge; the settled API rate is all-in.

Provider prices can change over time. The exact amount charged for each completed request is available in your dashboard.

Model catalog API

Use your API key to retrieve currently available models and their current catalog rates. A fallback route can change the final rate for a request, subject to the direct-list-price ceiling above. Filter by type, vision, video, reasoning, streaming, or provider to find models that meet a workload's requirements.

curl 'https://api.cheaperinference.com/v1/models?type=text&vision=true&streaming=true' \
  -H 'Authorization: Bearer ci_live_YOUR_API_KEY'

Token rates are USD per 1 million tokens. Fixed media rates use media_input_unit_price and media_unit_price per unit. For tiered models, use pricing.above_threshold when the request exceeds input_token_price_threshold.

Model prices do not change on a fixed customer-facing schedule: they can change when an upstream rate or an eligible route changes. Do not treat a locally cached rate as the final charge for a request. Use these response fields to validate a cache:

• pricing_version is a SHA-256 version of the returned, API-key-filtered catalog. An unchanged value means that catalog is unchanged.
• pricing_checked_at is the UTC time of the latest successful provider-price check.
• pricing_updated_at is the latest UTC catalog change among the returned models.

Pricing change feed

curl 'https://api.cheaperinference.com/v1/pricing/changes?since=2026-07-29T00:00:00Z' \
  -H 'Authorization: Bearer ci_live_YOUR_API_KEY'

since is a required RFC 3339 timestamp. The endpoint returns the latest current customer-facing model object for each catalog row changed at or after that time. A removal has change_type: "removed" and model: null. Follow next_cursor while has_more is true; keep the original since value on every page.

1{2  "object": "pricing.changes",3  "pricing_version": "sha256:8f4c...",4  "pricing_checked_at": "2026-07-29T14:32:10Z",5  "pricing_updated_at": "2026-07-29T14:31:52Z",6  "data": [{7    "model_id": "gpt-5.4",8    "change_type": "updated",9    "changed_at": "2026-07-29T14:31:52Z",10    "model": {11      "id": "gpt-5.4",12      "object": "model",13      "type": "text",14      "endpoint": "/v1/chat/completions",15      "pricing": {16        "currency": "USD",17        "input_per_million": "1.250000",18        "cache_read_input_per_million": "0.125000",19        "cache_write_input_per_million": "1.250000",20        "output_per_million": "10.000000"21      }22    }23  }],24  "has_more": false,25  "next_cursor": null26}

There is no pricing webhook at this time. Poll this feed, process changes idempotently by model_id and changed_at, then use the newest changed_at as the next since value. Because since is inclusive, the boundary change may be returned again. Periodically compare pricing_version with GET /v1/models to reconcile the full cache.

The catalog identifies modality, endpoint, vision, video, reasoning and streaming support. Tool calling, structured output, context limits, and accepted optional parameters can still vary by model and serving provider; verify the exact request shape before production use.

Parameters

"response_format": {
  "type": "json_object"
}

Price ceiling

Add min_discount_percent to a chat completions, completions, messages, or responses request. The value sets the minimum discount off list price that the request will accept.

The field takes a number from 0 to 99.99. The gateway checks the value against our published list price for the model. It never checks it against a price claimed by a supply source. The gateway removes the field from the request body before it forwards the request, so no upstream provider sees it.

1curl https://api.cheaperinference.com/v1/chat/completions \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "gpt-5.4",6    "messages": [{"role": "user", "content": "Hello!"}],7    "min_discount_percent": 408  }'

Only supply priced at or under the ceiling can serve the request. The gateway checks the input rate and the output rate separately. A source must clear the ceiling on both rates to qualify. The gateway also caps the bill for the request at list price times (1 − min_discount_percent / 100) for each rate, even when the serving source would allow a higher charge.

When no supply currently clears the ceiling, the request fails. The gateway does not fall back to a source priced above the ceiling.

Errors

• 400 / invalid_min_discount: min_discount_percent is not a number, or it is outside 0–99.99.
• 400 / list_price_unknown: the gateway has no list price for this model, so it cannot check a ceiling.
• 503 / min_discount_unavailable: no supply is currently priced at or under the ceiling. Lower min_discount_percent or retry later.

1{2  "error": {3    "message": "No supply for gpt-5.4 currently meets min_discount_percent=40 (a required discount of at least 40%; customer ceiling $1.80 in / $9.00 out per 1M tokens; upstream list $3.00 / $15.00). Lower min_discount_percent or retry later.",4    "type": "api_error",5    "param": null,6    "code": "min_discount_unavailable"7  }8}

Check available supply first

Use GET /v1/models/supply to count sources that currently clear a ceiling for a model, before you send a real request.

curl 'https://api.cheaperinference.com/v1/models/supply?model=gpt-5.4&min_discount_percent=40' \
  -H "Authorization: Bearer ci_live_YOUR_API_KEY"

1{2  "object": "model.supply",3  "model": "gpt-5.4",4  "min_discount_percent": 40,5  "list_input_per_million": "3.00",6  "list_output_per_million": "15.00",7  "max_input_per_million": "1.80",8  "max_output_per_million": "9.00",9  "supply_buffer_percent": 4,10  "candidate_count": 3,11  "candidates": ["primary", "network_a", "network_b"],12  "primary_market_offers": 2,13  "primary_market_min_discount_percent": 12,14  "note": "Candidates are priced under the ceiling now. Health and fill are not guaranteed."15}

candidate_count is the number of sources priced under the ceiling right now. max_input_per_million and max_output_per_million are the capped billing rates for this ceiling. primary_market_offers is null when the primary marketplace could not be checked for this request. A candidate count above zero reports current pricing only. It does not guarantee provider health or a successful fill.

For the full parameter and schema reference, see the API reference.

Zero data retention

Add zdr: true to a chat completions, completions, messages, or responses request. The gateway then routes the request only through providers that support a zero-data-retention policy. A zero-data-retention provider does not keep your prompts or responses and does not train on them.

1curl https://api.cheaperinference.com/v1/chat/completions \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "claude-sonnet-5",6    "messages": [{"role": "user", "content": "Hello!"}],7    "zdr": true8  }'

The gateway removes the field from the request body before it forwards the request. The flag adds to the Data Protection setting on your account. A request cannot switch that setting off with zdr: false.

A ZDR request excludes every provider that does not support a zero-data-retention policy. The pool is smaller, so the price is usually higher than the standard price for the same model. The catalog shows the ZDR price when you switch on the ZDR only toggle, and GET /v1/models?zdr=true returns the same view.

curl 'https://api.cheaperinference.com/v1/models?zdr=true' \
  -H "Authorization: Bearer ci_live_YOUR_API_KEY"

Errors

• 400 / invalid_zdr: zdr is not a boolean.
• 503 / zdr_capacity_unavailable: no provider with zero-data-retention support currently serves the model. Choose another model. If zdr came from the request body, you can also remove it; if it came from the account's Data Protection setting, change that setting instead.

Use GET /v1/models/supply?model=…&zdr=true to count the zero-data-retention sources for a model before you send a real request.

Ranking: speed, discount, balance

Several discounted sources can usually serve one request: the marketplace, seller networks, and direct providers. The gateway tries them in order until one succeeds, within a time limit. Only when none of them can serve the request in time does it fall back to a reference rail (OpenRouter or the Vercel AI Gateway). Add ranking to a chat completions, completions, messages, or responses request to choose the order of the discounted sources. The rails are never part of that ranking.

1curl https://api.cheaperinference.com/v1/chat/completions \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "gpt-5.4",6    "messages": [{"role": "user", "content": "Hello!"}],7    "stream": true,8    "ranking": "speed"9  }'

• "discount": the cheapest source first. This is the order every request used before this field existed.
• "speed": the source with the highest recent speed score first. Cost breaks ties.
• "balance" (default): speed score multiplied by discount, highest first.

The speed score comes from the gateway's own traffic on the same model, not from a fixed list of "fast" providers. For each discounted source it measures the time to the first visible output token and the output tokens per second over the last twelve hours, with the last hour weighted most. The score is relative to the other sources on that model. A source with fewer than five recent requests scores neutral, so a model with little traffic ranks by cost until the signal builds.

Speed and discount are not a trade-off. The fastest source for a model is often also one of the cheapest, so "speed" can pick the same route as "discount", and "discount" does not promise a slower answer. The field says how the gateway ranks the candidates given the recent telemetry. It does not add or remove candidates, and it does not change the price rule: the source that serves the request sets the price you pay.

The gateway removes the field from the request body before it forwards the request. Combine it freely with min_discount_percent and zdr: those two decide which sources are eligible, ranking decides the order among them.

Errors

• 400 / invalid_ranking: ranking is not one of speed, discount, balance.

Images and vision

Runnable vision request

Use an HTTPS image URL or a base64 data URL in an OpenAI image_url content part:

1curl https://api.cheaperinference.com/v1/chat/completions \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "gpt-5.4",6    "messages": [{7      "role": "user",8      "content": [9        {"type": "text", "text": "Describe this image."},10        {11          "type": "image_url",12          "image_url": {"url": "https://example.com/image.png"}13        }14      ]15    }]16  }'

Generate an image

Image-generation models use POST /v1/images/generations, not Chat Completions. The response contains data entries with either a URL or base64 image data.

1curl https://api.cheaperinference.com/v1/images/generations \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "nano-banana",6    "prompt": "A tiny observatory on a mossy hill at blue hour",7    "n": 18  }'

Edit an image

Use POST /v1/images/edits only with a model whose capabilities.image_edit value is true in GET /v1/models. Multipart requests accept one to ten PNG, JPEG, or WebP source images and an optional PNG mask; decoded sources plus the mask may total at most 20 MiB. An unsupported edit is refused and is never converted into a text-to-image generation.

1curl https://api.cheaperinference.com/v1/images/edits \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -F "model=gpt-image-2" \4  -F "prompt=Replace the background with a quiet library" \5  -F "image=@source.png;type=image/png" \6  -F "mask=@mask.png;type=image/png"

The JSON extension accepts source HTTP(S) URLs in input_references. Cheaper Inference passes those URLs only to compatible providers and never downloads them itself. Masks must remain PNG uploads or PNG data URLs so their semantics cannot be silently dropped by a provider route.

Generate a video

Video models use POST /v1/videos/generations. Generation is buffered while the provider completes the job, so use a client timeout of at least five minutes. The response contains avideos array with URLs or base64 MP4 data.

1curl https://api.cheaperinference.com/v1/videos/generations \2  -H "Authorization: Bearer ci_live_YOUR_API_KEY" \3  -H "Content-Type: application/json" \4  --max-time 300 \5  -d '{6    "model": "seedance-2.0",7    "prompt": "A paper boat drifting through a neon-lit rain puddle",8    "duration": 5,9    "resolution": "720p",10    "n": 111  }'

Audio and voice: This API does not currently expose speech generation, transcription, or voice endpoints, and the model catalog contains no audio-generation rows. Do not send audio models to Chat Completions; audio examples will be added when a billable audio route is available.

Prompt caching

cache_control is passed through when supported by the selected model. Cache reads and writes use imported model-specific rates when available. If no cache-read rate is available, cached reads use 10% of the normal input rate; cache writes without a separate rate use the normal input rate.

cURL Anthropic Messages request

Available to all users: every Cheaper Inference account can use these endpoints with an active API key that has Inference scope. No Anthropic account or separate enablement is required. Normal wallet balance, model access, and API-key limits still apply.

Compatibility scope: this integration implements POST /v1/messages and POST /v1/messages/count_tokens. It is not the complete Anthropic platform API; unsupported features are listed below.

1curl https://api.cheaperinference.com/v1/messages \2  -H "X-Api-Key: ci_live_YOUR_API_KEY" \3  -H "Anthropic-Version: 2023-06-01" \4  -H "Content-Type: application/json" \5  -d '{6    "model": "claude-opus-4.8",7    "max_tokens": 1024,8    "messages": [{"role": "user", "content": "Hello!"}]9  }'

1from anthropic import Anthropic23client = Anthropic(4    api_key="ci_live_YOUR_API_KEY",5    base_url="https://api.cheaperinference.com",6)78message = client.messages.create(9    model="claude-opus-4.8",10    max_tokens=1024,11    messages=[{"role": "user", "content": "Hello!"}],12)

POST /v1/messages accepts the Anthropic Messages shape and returns an Anthropic-style message. It supports buffered and streamed text, system content, vision blocks, client-side tools and tool results, multi-turn replay, stop sequences, tool choice, usage accounting, and extended-thinking request hints. Streams use Anthropic event names such as message_start, content_block_delta, message_delta, and message_stop.

Use POST /v1/messages/count_tokens with the same model, system, messages, and tools fields for an estimated input_tokens count. Counting does not submit an inference request or charge the wallet.

Client-side tools work because Claude Code executes them locally and returns tool_result blocks. Anthropic-hosted server tools, Message Batches, Files, and signed thinking-block replay are not compatibility targets. Thinking settings can guide an eligible model, but responses do not expose Anthropic-signed thinking blocks.

For the complete environment setup and model mapping, use the Claude Code integration guide.

Errors and limits

Common errors

OpenAI-compatible endpoints use the envelope below. The Messages endpoints return Anthropic's {"type":"error","error":{...}} shape instead. Use the error type for program logic and keep the message for logs and operator context.

1{2  "error": {3    "message": "Invalid API key.",4    "type": "authentication_error",5    "param": null,6    "code": "invalid_api_key"7  }8}

• 400 / invalid_request: unsupported model, endpoint, or invalid request body. Check the JSON body, model id, and required fields.
• 401 / invalid_api_key: invalid or missing API key. Create or rotate a key, then update the Bearer token.
• 402 / insufficient_balance: add funds in Billing, then retry the request.
• 403 The API key does not allow the model, client IP, or request.
• 404 / request_not_found: feedback can only target a request on your own account.
• 413 Request body or vision payload is too large.
• 422 A query or typed request parameter failed validation.
• 429 / rate_limit_error: a rate, concurrency, or quota limit was reached. Retry with exponential backoff and jitter, and respect the Retry-Afterresponse header before retrying.
• 502 / upstream_error: provider or transport failure after eligible routes were exhausted. The request is recorded but not charged.
• 503 No eligible provider route is currently available.
• 504 The serving provider timed out.

Reliability

1{2  "usage": {3    "prompt_tokens": 1234,4    "completion_tokens": 321,5    "total_tokens": 1555,6    "cost": 0.012345,7    "cost_details": {8      "upstream_inference_cost": 0.0123459    }10  },11  "cheaper_inference": {12    "request_id": 4812,13    "billing": {14      "status": "settled",15      "billed_cost_usd": "0.012345",16      "currency": "USD"17    }18  }19}

usage.cost and usage.cost_details.upstream_inference_cost both report the exact all-in customer charge after settlement. From the caller's perspective Cheaper Inference is the upstream inference service; the private amount paid to the selected serving provider is not exposed. cheaper_inference.billing.billed_cost_usd is the persisted amount deducted for that request after routing and settlement. For Chat Completions and Completions, streaming sends this metadata in a final empty-choice event immediately before [DONE]. Responses API streams include it in the final response.completed object. Messages streams include it on the final message_delta event. A failed or unsettled request does not report a billed amount.

Cheaper Inference does not store prompt or response bodies in its application database. Prompts are forwarded to the provider that serves the request, and provider-side handling is subject to that provider's applicable terms. Prompt caching may also create provider-side cached state. Request metadata needed for billing and operations is retained, including model, endpoint, token counts, charged amount, status, and a sanitized failure reason. Review the privacy policy for the full statement.

Providers and request shapes

Every request is served by one of several supply sources: a marketplace, seller networks that bring their own provider credentials, direct providers on our credentials, and benchmark rails. Before ranking them by cost, the router removes any source that does not honour the shape of your request. A shape is the set of features a request uses: the endpoint, streaming, tools and a forced tool_choice, a reasoning effort, JSON mode, image input, logprobs.

The table lists every shape a source is known not to honour, from live probes we re-run when a provider changes. A source that is not listed for a shape honours it on every model we tested. When no source is left for a shape, the request is still served on the cheapest eligible source and the affected field may come back empty.

Two shapes are surface-bound rather than source-bound: Anthropic Messages requests that carry signed thinking blocks are served only by sources with a native Messages surface, and namespaced tool choices on /v1/responses only by sources with a native Responses surface. Model capabilities (vision, reasoning, video) come from the model catalog and are not repeated here.

Usage and billing API

Usage reporting is workspace-wide and accepts the same API-key authentication as inference. Create a key with Usage reporting only access in API Keys when the reporting system should not be able to submit inference requests or spend wallet funds. Add api_key_id to restrict either endpoint to one key.

Request history

curl 'https://api.cheaperinference.com/v1/usage/requests?start_at=2026-07-01T00:00:00Z&end_at=2026-08-01T00:00:00Z&limit=100' \
  -H 'Authorization: Bearer ci_live_YOUR_API_KEY'

• start_at and end_at are optional RFC 3339 timestamps. The start is inclusive and the end is exclusive.
• api_key_id optionally filters to a key owned by the same workspace.
• limit accepts 1–100 and defaults to 100.
• cursor accepts the previous page's next_cursor. Results are newest first.

1{2  "object": "list",3  "scope": "workspace",4  "data": [{5    "request_id": 4812,6    "api_key_id": 19,7    "api_key_name": "production",8    "model": "gpt-5.4",9    "endpoint": "/v1/chat/completions",10    "status": "settled",11    "prompt_tokens": 1234,12    "completion_tokens": 321,13    "cache_read_input_tokens": 0,14    "cache_write_input_tokens": 0,15    "total_tokens": 1555,16    "billed_cost_usd": "0.012345",17    "currency": "USD",18    "total_latency_ms": 1890,19    "time_to_first_token_ms": 420,20    "created_at": "2026-07-29T14:30:00Z",21    "completed_at": "2026-07-29T14:30:02Z"22  }],23  "has_more": false,24  "next_cursor": null25}

billed_cost_usd is non-null only when status is settled. Request history intentionally excludes prompts, generated content, provider cost, private route names, and profit or margin data.

Daily spend

curl 'https://api.cheaperinference.com/v1/usage/daily?start_at=2026-07-01T00:00:00Z&end_at=2026-08-01T00:00:00Z' \
  -H 'Authorization: Bearer ci_live_YOUR_API_KEY'

• When dates are omitted, the endpoint returns the preceding 30 days through the current time.
• Every calendar date in the range is returned, including dates with zero requests.

1{2  "object": "usage.daily",3  "scope": "workspace",4  "currency": "USD",5  "start_at": "2026-07-01T00:00:00Z",6  "end_at": "2026-08-01T00:00:00Z",7  "total_requests": 820,8  "settled_requests": 811,9  "spend_usd": "148.420015",10  "prompt_tokens": 1043200,11  "cached_tokens": 786400,12  "cache_reported_request_count": 811,13  "cache_hit_pct": 75.4,14  "daily_spend": [{15    "date": "2026-06-30",16    "request_count": 24,17    "settled_request_count": 24,18    "spend_usd": "3.910042",19    "prompt_tokens": 30500,20    "cached_tokens": 22100,21    "cache_reported_request_count": 24,22    "cache_hit_pct": 72.523  }]24}

Daily spend sums persisted charges for settled requests only. Therefore, total_requests can be greater than settled_requests.

Balance

curl 'https://api.cheaperinference.com/v1/account/balance' \
  -H 'Authorization: Bearer ci_live_YOUR_API_KEY'

1{2  "object": "account.balance",3  "currency": "USD",4  "balance_usd": 124.50,5  "available_usd": 118.02,6  "reserved_usd": 6.48,7  "auto_recharge_enabled": true,8  "threshold_usd": 25.00,9  "recharge_amount_usd": 100.00,10  "workspace_name": "Acme",11  "funded_by_owner": false12}

Spend against available_usd, not balance_usd. The difference is reserved_usd: money already held against requests that have not settled yet, so the balance on its own overstates what can still be spent. This endpoint needs the account:read scope — a usage-only key receives a 403.

API-key access

• inference permits inference requests.
• usage:read permits request history, daily spend, and pricing-change reads.
• account:read permits balance and account reads.
• GET /v1/models accepts any valid API key.
• Use a reporting-only key for billing exports and monitoring systems.

Cursor compatibility

Cursor Ask and Agent can use OpenAI-family models through the OpenAI base-URL override. In Cursor Settings → Models, add a dedicated Cheaper Inference key, enable Override OpenAI Base URL, and enter https://api.cheaperinference.com/v1. Select an exact model ID from the live catalog, then verify both a short Ask prompt and one harmless Agent edit.

Cursor sends Agent tool traffic through Chat Completions. The gateway normalizes the current streamed ApplyPatch tool shape and the older Responses-shaped BYOK request format. Cursor's override is global for OpenAI-family models; tab completion, Claude- and Gemini-family models, and other Cursor-hosted features do not use this connection.

Use key scope, model, rate, spend, and expiration controls. Do not restrict the key to your workstation IP because Cursor sends these requests from its own service. See the complete Cursor setup and troubleshooting guide.

Production checklist

• Load your API key from a server-side environment variable.
• Use GET /v1/models to confirm the model is available and supports the required modality.
• Set an explicit client timeout appropriate for long model generations.
• For direct HTTP clients, retry transport errors, 429, and 5xx responses with exponential backoff.
• Do not retry 400, 401, 402, or 413 without correcting the request. For 402, add funds before retrying.
• Treat an interrupted stream as incomplete and retry the entire request.
• Review token usage, spend, and failures in the dashboard before increasing traffic.

For the generated public request, response, authentication, and error schemas, open the API reference.