Classify Comment - Developer Platform

Classify a comment in one API call. Returns intent labels, sentiment, detected language and Arabic dialect, a suggested reply in the commenter’s own dialect, and the reasoning behind the classification.

Cost: $0.006 per request (6 credits). Semantic cache hits are free (X-NAWA-Cache: HIT) and do not decrement your credit balance.

Request

Headers

Header	Required	Description
`Authorization`	Yes	`Bearer nawa_live_sk_xxx` or `Bearer nawa_test_sk_xxx`
`Content-Type`	Yes	`application/json`

Body

Parameter	Type	Required	Description
`text`	string	Yes	The comment text to classify. Must be a non-empty string.

Query parameters

Parameter	Type	Required	Description
`provider`	string	No	Force a specific AI provider for A/B testing. One of `claude`, `gemini`. Default: chosen by language routing.

Example request

curl -X POST https://api.trynawa.com/v1/classify \
  -H "Authorization: Bearer nawa_test_sk_xxx" \
  -H "Content-Type: application/json" \
  -d '{"text": "متى الجزء الثاني؟"}'

import { Nawa } from '@nawalabs/sdk'

const nawa = new Nawa({ apiKey: process.env.NAWA_API_KEY })

const { data, error } = await nawa.classify({
  text: 'متى الجزء الثاني؟',
})

from nawa import Nawa

nawa = Nawa(api_key="your_api_key")

result = nawa.classify(text="متى الجزء الثاني؟")

Response

Success response (200)

{
  "result": {
    "id": "cls_nw_lx4cqy2j2o1o",
    "object": "classification",
    "intent": ["question"],
    "sentiment": "neutral",
    "language": "ar",
    "dialect": "gulf",
    "dialect_confidence": 0.95,
    "requires_response": true,
    "priority": "medium",
    "suggested_reply": {
      "text": "يعطيك العافية يا غالي 🙏 الجزء الثاني قريب إن شاء الله، خلك متابع عشان ما يفوتك!",
      "direction": "rtl"
    },
    "reasoning": "The comment asks when the next part is released, which needs a direct answer rather than a reaction.",
    "provider": "claude",
    "model": "claude-haiku-4-5-20251001",
    "fallback_used": false,
    "tokens_used": 988,
    "cost_usd": 0.006,
    "credits_used": 6
  },
  "success": true,
  "errors": [],
  "request_id": "req_nw_9cfb228162d6f8c3"
}

Result fields

Field	Type	Description
`id`	string	Classification ID in `cls_nw_xxx` format
`object`	string	Always `"classification"`
`intent`	string[]	One or more intent labels. Typical values: `question`, `complaint`, `praise`, `suggestion`, `spam`, `other`. The array may contain multiple labels when a comment expresses more than one intent.
`sentiment`	string	One of `positive`, `negative`, `neutral`, `mixed`
`language`	string	Detected language: `ar`, `en`, or `mixed`
`dialect`	string \| null	Detected Arabic dialect: `gulf`, `egyptian`, `levantine` or `msa`. `null` for non-Arabic text.
`dialect_confidence`	number \| null	Confidence in the detected dialect, 0 to 1. `null` for non-Arabic text.
`requires_response`	boolean	Whether the model judges this comment needs a reply
`priority`	string	One of `low`, `medium`, `high`
`suggested_reply.text`	string	A draft reply, written in the commenter’s own language and dialect. Empty string when the classification decided the comment needs a reaction rather than a reply.
`suggested_reply.direction`	string	`"rtl"` for Arabic, `"ltr"` otherwise. Useful for UI rendering.
`reasoning`	string \| null	Why the classifier chose this action. Useful for review queues and for tuning your own thresholds.
`provider`	string	AI provider that produced this result: `claude` or `gemini`
`model`	string	The model that actually ran, e.g. `claude-haiku-4-5-20251001`. Classification runs on Haiku for cost; reply generation may use a larger model.
`fallback_used`	boolean	`true` if the primary provider failed and a fallback produced this result
`tokens_used`	number \| null	Input plus output tokens across classification and reply generation. `null` only when no model call was made.
`cost_usd`	number	Always `0.006` for this endpoint
`credits_used`	integer	Always `6` for this endpoint

Changed 2026-09-13. Three fields on this endpoint used to be placeholders and are now real: dialect was hardcoded to "gulf" for all Arabic, dialect_confidence returned the classifier’s overall confidence rather than a dialect score, and suggested_reply.text returned internal reasoning instead of a reply. If you built around any of those, re-read the table above. The reasoning moved to its own reasoning field.

Response headers

Header	Description
`X-Request-Id`	Unique request identifier. Quote this to support when reporting an issue.
`X-NAWA-Cache`	`HIT` if served from the semantic cache (free, no credit charge), `MISS` otherwise.
`X-NAWA-Provider`	Which provider produced the response: `claude` or `gemini`.
`X-NAWA-Latency`	Server-side latency in milliseconds.
`X-NAWA-Calibration-Version`	Calibration threshold version applied to this response. `"default"` when no custom calibration is set.
`X-NAWA-Environment`	`sandbox` for free keys, `live` for paid keys.
`X-RateLimit-Limit`	Your tier’s per-minute request ceiling.
`X-RateLimit-Remaining`	Requests remaining in the current one-minute window.
`X-RateLimit-Reset`	RFC 3339 timestamp when the window resets.

Error responses

Status	Type	When
400	`invalid_request_error`	Missing or empty `text`, invalid `provider` query param, malformed JSON body
401	`authentication_error`	Missing, malformed, revoked, or expired API key
402	`insufficient_credits`	Live key has insufficient credit balance for this request. Response body includes `balance_credits` and `required_credits` on the error object.
429	`rate_limit_error`	Per-minute rate limit exceeded, or sandbox lifetime quota (100 requests) exhausted
500	`api_error`	Unexpected internal failure. The response `message` is a generic constant; quote `request_id` to support for the underlying cause.

See Errors for the full envelope shape and all error codes.

More examples

English praise

Request:

curl -X POST https://api.trynawa.com/v1/classify \
  -H "Authorization: Bearer nawa_test_sk_xxx" \
  -H "Content-Type: application/json" \
  -d '{"text": "This is hands down the best review I have seen on this phone. Subscribed!"}'

Response:

{
  "result": {
    "id": "cls_nw_en01praise",
    "object": "classification",
    "intent": ["praise"],
    "sentiment": "positive",
    "language": "en",
    "dialect": null,
    "dialect_confidence": null,
    "requires_response": false,
    "priority": "low",
    "suggested_reply": {
      "text": "Highly positive engagement from a new subscriber. No response required; consider pinning or hearting for social signal.",
      "direction": "ltr"
    },
    "provider": "claude",
    "model": "claude-sonnet-4-5-20250929",
    "fallback_used": false,
    "tokens_used": null,
    "cost_usd": 0.006,
    "credits_used": 6
  },
  "success": true,
  "errors": [],
  "request_id": "req_nw_en01praise"
}

English complaint (multi-intent)

{
  "result": {
    "id": "cls_nw_en02complaint",
    "object": "classification",
    "intent": ["complaint", "suggestion"],
    "sentiment": "negative",
    "language": "en",
    "dialect": null,
    "dialect_confidence": null,
    "requires_response": true,
    "priority": "medium",
    "suggested_reply": {
      "text": "Audio quality complaint with an implicit suggestion to fix. Acknowledge and commit to the next upload being better.",
      "direction": "ltr"
    },
    "provider": "claude",
    "model": "claude-sonnet-4-5-20250929",
    "fallback_used": false,
    "tokens_used": null,
    "cost_usd": 0.006,
    "credits_used": 6
  },
  "success": true,
  "errors": [],
  "request_id": "req_nw_en02complaint"
}

Arabic question

{
  "result": {
    "id": "cls_nw_ar01question",
    "object": "classification",
    "intent": ["question"],
    "sentiment": "neutral",
    "language": "ar",
    "dialect": "gulf",
    "dialect_confidence": 0.88,
    "requires_response": true,
    "priority": "medium",
    "suggested_reply": {
      "text": "Viewer asks when the next part drops. Reply with the schedule in the same dialect.",
      "direction": "rtl"
    },
    "provider": "claude",
    "model": "claude-sonnet-4-5-20250929",
    "fallback_used": false,
    "tokens_used": null,
    "cost_usd": 0.006,
    "credits_used": 6
  },
  "success": true,
  "errors": [],
  "request_id": "req_nw_ar01question"
}

​Request

​Headers

​Body

​Query parameters

​Example request

​Response

​Success response (200)

​Result fields

​Response headers

​Error responses

​More examples

Request

Headers

Body

Query parameters

Example request

Response

Success response (200)

Result fields

Response headers

Error responses

More examples