Perplexity's Gateway API catalog expands to 5 models with 2 NVIDIA additions
Perplexity added two NVIDIA Nemotron models to its week-old Gateway API — nemotron-3.5-lightning-30b-a3b, now the catalog's cheapest model at $0.0115/1M input tokens, and nemotron-3-ultra-550b-a55b — growing the Perplexity-hosted open-weight catalog from 3 to 5 models.
The Gateway API, launched 2026-08-11, offered 3 Perplexity-hosted open-weight models: perplexity/deepseek-v4-flash-0731 ($0.13/$0.26/$0.028 per 1M input/output/cache-read tokens), perplexity/kimi-k3 ($3.00/$15.00/$0.30), and perplexity/glm-5.2 ($1.40/$4.40/$0.14).
By 2026-08-14 the catalog listed 5 models, adding perplexity/nemotron-3.5-lightning-30b-a3b ($0.0115/$0.17/$0.00115) and perplexity/nemotron-3-ultra-550b-a55b ($0.25/$2.50/$0.25). The 3 original models' prices are unchanged.
Proof of change
Perplexity AI's pricing pages, as we captured them on two dates.
These values appear in only one of the two captures. That can mean a page-layout difference rather than a price move — read the images, not just the list.
Showing the whole page as captured — scroll either panel, or open it at full size.
Show 3 other pages we compared
Showing the whole page as captured — scroll either panel, or open it at full size.
Showing the whole page as captured — scroll either panel, or open it at full size.
Showing the whole page as captured — scroll either panel, or open it at full size.
- 1 page had no counterpart in the earlier capture and is not shown.
Both images are our own captures, taken on the dates shown. The pixels are unmodified — nothing is retouched, and where a region is highlighted the marker is drawn over the image, not into it. Open either capture to see it at full resolution.
Three days after launching its Gateway API — Perplexity’s first developer product priced with real hosting margin rather than at-cost third-party resale — the company expanded the model catalog from 3 to 5. Both additions are NVIDIA-origin: perplexity/nemotron-3.5-lightning-30b-a3b, which at $0.0115 input / $0.17 output per 1M tokens undercuts every other model in the catalog (including the previous cheapest, DeepSeek’s $0.13/$0.26), and perplexity/nemotron-3-ultra-550b-a55b at $0.25/$2.50.
Both models are reachable through the same infrastructure as the original three — an OpenAI-compatible Chat Completions endpoint and an Anthropic-compatible Messages endpoint under one API key, with the live GET /models catalog doubling as the request allowlist. No price moved on the three original Gateway models (DeepSeek, Kimi, GLM), and no other API surface (Sonar, Search API, Agent API, Embeddings) was affected.
Perplexity's Agent API `fetch_url` tool price doubled back to $0.0005 per invocation, undoing the halving to $0.00025 that took effect on 2026-07-29. Confirmed by two independent worked cost examples on the docs page: the "Agent API Research Preset" total is $0.007 on 2026-08-14 versus $0.00675 on 2026-08-11 — a difference that only reconciles if fetch_url returned to $0.0005. `web_search` ($0.0025), `people_search`/`finance_search` ($0.005), and the $0.03 `sandbox` per-session fee are all unchanged, as are Sonar, Search API, Gateway, and Embeddings rates.