fix: add Perplexity models sonar-deep-research

[github-actions]

fix: add Perplexity models sonar-deep-research

Closes #673

Source issue: #673

Summary

Field	Value
Provider	perplexity
Primary model	sonar-deep-research
Changed models	sonar-deep-research
Added models	sonar-deep-research
Updated models	None
Verification sources	1 2 3 4

Verified metadata

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
sonar-deep-research	Sonar Deep Research		perplexity	openai	chat	input=128000, output=not provided	in/out=2/8 per 1M	active

Verification notes

Verification

Official source verification

• https://docs.perplexity.ai/docs/sonar/models/sonar-deep-research: Confirmed context length of 128K, model description as "expert-level research model conducting exhaustive searches and generating comprehensive reports", input cost $2/1M tokens, output cost $8/1M tokens.
• https://docs.perplexity.ai/api-reference/sonar-post: Confirmed sonar-deep-research is listed as a supported model enum value alongside sonar, sonar-pro, and sonar-reasoning-pro. The max_tokens parameter supports up to 128,000.
• https://docs.perplexity.ai/pricing: Confirmed pricing: input $2/1M, output $8/1M, citation tokens $2/1M, reasoning tokens $3/1M, search queries $5/1K.
• https://docs.perplexity.ai/guides/model-cards: Confirmed model exists and is categorized as a research model.

Fields verified

• format: "openai" — consistent with all existing Perplexity sonar models in the catalog
• flavor: "chat" — consistent with all existing Perplexity sonar models
• displayName: "Sonar Deep Research" — follows the pattern of "Sonar Pro", "Sonar Reasoning Pro"
• input_cost_per_mil_tokens: 2 — confirmed by pricing page and model docs page
• output_cost_per_mil_tokens: 8 — confirmed by pricing page and model docs page
• max_input_tokens: 128000 — confirmed by model docs page ("128K context length"), consistent with sonar, sonar-reasoning, sonar-reasoning-pro
• available_providers: ["perplexity"] — confirmed by API reference listing the model on the sonar endpoint

Fields omitted (not published or not applicable)

• max_output_tokens: Not documented as a separate limit by Perplexity. Omitted, consistent with sonar, sonar-reasoning, and sonar-reasoning-pro entries.
• parent: Not applicable — this is a standalone model, not a snapshot or variant of another model.
• reasoning: Not set. While the model uses reasoning tokens internally, existing sonar-reasoning and sonar-reasoning-pro entries also do not set this flag.
• multimodal: Not documented by Perplexity for this model.
• deprecated / deprecation_date: Not applicable — model is active.
• supported_regions: Not applicable — Perplexity is not a Vertex provider.
• Cache pricing fields: Not published by Perplexity.

sync_models (LiteLLM) cross-check

The model sonar-deep-research does not exist in sync_models (model_prices_and_context_window_backup.json). No cross-check is possible. All proposed values are sourced directly from official Perplexity documentation.

sync_models vs proposed update

sync_models cross-check found no pricing/token discrepancies for the changed models.

No sync_models discrepancies were noted.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	May 29, 2026 1:15pm