Grass Launches Contents API, a Live Web Retrieval Tool for AI Agents

Grass GRASS$0.646+3.2%, the bandwidth-sharing network whose GRASS token trades on Solana, launched the Grass Contents API on 8 October 2026. Grass describes the product as "an API that turns any public URL into clean, model-ready HTML, Markdown, or text", and in its launch thread called it "our first Live Context Retrieval tool for agents". Access is by request, and the announcement gives no pricing.

What the Contents API does for AI agents

The Grass Contents API takes a public web address and returns the page in a format a language model can read. An AI agent answering a question about today's prices, listings or news has to read live pages, and Grass says much of the web is difficult for agents to access reliably.

According to Grass, each request moves through up to three retrieval paths. Simple pages get a lightweight fetch meant to return quickly. If that fails, the request falls back to what Grass calls its "proprietary retrieval infrastructure". Pages that still do not load are opened in a full browser environment, which runs the page's JavaScript and returns the result.

Grass also lists geo-targeted requests (asking for a page as seen from a chosen location), freshness controls, and automatic removal of unneeded HTML so a model spends fewer tokens, the units of text AI providers bill by, on each page.

Why residential bandwidth matters for web retrieval

The Contents API is "built using the Grass Network", Grass wrote, and "the residential bandwidth of the Grass Network gives exceptional access to agent requests for live data". The Grass Network is made of people who install the Grass app and share their unused home internet connection, a model crypto calls DePIN, short for decentralized physical infrastructure network. The Grass website claims more than 8.5 million users, who earn points for staying connected and for traffic their connection handles. Grass says those points count toward rewards.

The launch post and thread do not mention Solana, the GRASS token or any change to contributor rewards. The Solana connection is the network's token. Grass has not described an on-chain role for the Contents API, or said how revenue from the API reaches the people supplying bandwidth.

Grass benchmark: 98% on hard domains against Firecrawl, Parallel and Exa

The performance figures come from a test Grass ran itself, and no independent replication has been published. Grass says it sent requests to 100 public web destinations on 1 October 2026, covering e-commerce, social platforms, news, Cloudflare-protected sites and JavaScript-heavy pages, with caching switched off wherever a provider allowed it.

On that full set, Grass reports that it retrieved content from 97% of URLs, compared with 84% for Firecrawl, 76% for Parallel and 66% for Exa. On what Grass calls its "hard-domain test set", it reports 98% against 71% for the next-best provider.

How developers request access at lcr.net

Developers register for the Contents API at lcr.net. The Grass post says "requests will be reviewed to confirm all requirements for access are met", without listing those requirements. Current plans support between 2 and 20 requests per second, with more dedicated capacity for custom deployments, according to the post.

The launch follows a roadmap reported by Cryptobriefing, which wrote that chief executive Andrej Radonjic set out a roadmap on 28 September 2026 moving Grass from batch sales of AI training data toward live retrieval. Radonjic discussed the Grass web-scraping model in a 2024 Lightspeed interview.