[BOT ISSUE] Perplexity: add missing perplexity/kimi-k3 and perplexity/glm-5.2 Gateway models

[github-actions]

Gap description

Perplexity's Gateway API serves two open-weight models under provider-specific IDs that are missing from the catalog. The underlying models exist in the catalog through other providers, but the Perplexity-specific IDs (perplexity/ prefix) are absent.

Missing Models

Model ID	Underlying Catalog Entry	Perplexity Pricing (input/output/cached $/1M)
perplexity/kimi-k3	moonshotai/Kimi-K3 (line 10670, baseten+together)	$3.00 / $15.00 / $0.30
perplexity/glm-5.2	zai-org/GLM-5.2 (line 10606, baseten+together)	$1.40 / $4.40 / $0.26

Pricing matches the underlying catalog entries exactly.

Context

• Perplexity's Gateway API (api.perplexity.ai/router/v1/) provides OpenAI-compatible access to third-party models using perplexity/ prefixed IDs
• This follows the established catalog pattern where provider-specific IDs get separate entries (e.g., accounts/fireworks/models/kimi-k3 for Fireworks, x-ai/grok-4.20-multi-agent for OpenRouter)
• Issue [BOT ISSUE] Add perplexity/sonar (Perplexity Agent API) #1020 (open) covers the related perplexity/sonar Agent API model

Suggested Catalog Entries

"perplexity/kimi-k3": {
  "format": "openai",
  "flavor": "chat",
  "multimodal": true,
  "input_cost_per_mil_tokens": 3,
  "output_cost_per_mil_tokens": 15,
  "input_cache_read_cost_per_mil_tokens": 0.3,
  "displayName": "Perplexity: Kimi K3",
  "reasoning": true,
  "max_input_tokens": 1048576,
  "available_providers": ["perplexity"]
}

"perplexity/glm-5.2": {
  "format": "openai",
  "flavor": "chat",
  "input_cost_per_mil_tokens": 1.4,
  "output_cost_per_mil_tokens": 4.4,
  "input_cache_read_cost_per_mil_tokens": 0.26,
  "displayName": "Perplexity: GLM 5.2",
  "reasoning": true,
  "max_input_tokens": 262144,
  "max_output_tokens": 128000,
  "available_providers": ["perplexity"]
}

Verification Checklist

Check	Status	Detail
Cross-source	✅	Both models confirmed on Perplexity Gateway quickstart (model IDs in /router/v1/models response) AND Perplexity pricing page (pricing in Gateway API section)
Already in catalog	✅	grep for perplexity/kimi-k3 and perplexity/glm-5.2 in model_list.json returns no matches
ID format	✅	Follows perplexity/ prefix convention used by Gateway API; consistent with how OpenRouter-specific IDs use provider prefixes (e.g., x-ai/, moonshotai/)
Pricing match	✅	Perplexity prices match existing catalog entries for underlying models exactly
Not deprecated	✅	Active on Gateway API docs and pricing page
Not embedding/vision-only	✅	Both are chat models (kimi-k3 is multimodal, glm-5.2 is text-only with reasoning)
Duplicate check	✅	No open or closed issues cover perplexity/kimi-k3 or perplexity/glm-5.2

Verification Notes

Field	Source	Notes
Model IDs	Gateway quickstart	Listed in /router/v1/models API response
Pricing	Perplexity pricing	Gateway API section, per 1M tokens
Specs (kimi-k3)	Derived from moonshotai/Kimi-K3 catalog entry	Same underlying model, pricing matches
Specs (glm-5.2)	Derived from zai-org/GLM-5.2 catalog entry	Same underlying model, pricing matches

Relationship to existing entries

Entry	Providers	Status
moonshotai/Kimi-K3	baseten, together	Already in catalog (line 10670)
accounts/fireworks/models/kimi-k3	fireworks	Already in catalog
perplexity/kimi-k3	perplexity	Missing — this issue
zai-org/GLM-5.2	baseten, together	Already in catalog (line 10606)
perplexity/glm-5.2	perplexity	Missing — this issue

Local Files Inspected

• packages/proxy/schema/model_list.json — confirmed both IDs absent; verified underlying model entries and pricing

{
  "kind": "missing_model",
  "provider": "perplexity",
  "models": ["perplexity/kimi-k3", "perplexity/glm-5.2"],
  "status": "active",
  "model_specs": {
    "perplexity/kimi-k3": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 3,
      "output_cost_per_mil_tokens": 15,
      "input_cache_read_cost_per_mil_tokens": 0.3,
      "displayName": "Perplexity: Kimi K3",
      "reasoning": true,
      "max_input_tokens": 1048576,
      "available_providers": ["perplexity"]
    },
    "perplexity/glm-5.2": {
      "format": "openai",
      "flavor": "chat",
      "input_cost_per_mil_tokens": 1.4,
      "output_cost_per_mil_tokens": 4.4,
      "input_cache_read_cost_per_mil_tokens": 0.26,
      "displayName": "Perplexity: GLM 5.2",
      "reasoning": true,
      "max_input_tokens": 262144,
      "max_output_tokens": 128000,
      "available_providers": ["perplexity"]
    }
  },
  "source_urls": [
    "https://docs.perplexity.ai/docs/gateway/quickstart",
    "https://docs.perplexity.ai/guides/pricing"
  ]
}

[github-actions]

Prepared catalog updates for perplexity/kimi-k3, perplexity/glm-5.2. Added models: perplexity/kimi-k3, perplexity/glm-5.2.

Verification sources: https://docs.perplexity.ai/docs/gateway/quickstart, https://docs.perplexity.ai/guides/pricing
Verified metadata applied:

• perplexity/kimi-k3: format=openai, flavor=chat, displayName=Perplexity: Kimi K3, max_input_tokens=1048576, max_output_tokens=not provided, pricing(input/output per 1M)=3/15, input_cache_read_cost_per_mil_tokens=0.3, providers=perplexity, reasoning=true, multimodal=true
• perplexity/glm-5.2: format=openai, flavor=chat, displayName=Perplexity: GLM 5.2, max_input_tokens=262144, max_output_tokens=128000, pricing(input/output per 1M)=1.4/4.4, input_cache_read_cost_per_mil_tokens=0.26, providers=perplexity, reasoning=true

Included in the daily batch PR: #1070