Octolens alternative: a social data API, not a mention feed

Octolens is an AI mention feed. You add keywords in its app, it collects matching posts from the sources on your plan, scores them for relevance and sentiment, and its API (POST /api/v2/mentions) pages through what it collected, at up to 500 requests an hour per organisation. Social Fetch is the data layer underneath a feed like that: live keyword search, profiles, posts, comments and transcripts across 23 platforms, including Instagram, Threads and Facebook, which Octolens doesn't cover. Monitors poll the search for you and send new posts to your endpoint as signed webhooks. You pay per request, so a busy keyword doesn't raise the bill, and you run relevance filtering with your own model.

Octolens's API is built around mentions. POST /api/v2/mentions pages through posts that matched the keywords you set up, up to 100 per call, each with a body, author, source, relevance label, sentiment and tags. You can filter by source, sentiment, keyword, language, tag, engagement and date. Everything it returns traces back to a keyword you track.

That's the right shape for an alert feed. It stops when you need what sits around a mention: the author's other posts, the rest of the thread, what a TikTok video actually says. Social Fetch covers those as separate calls with the same key: GET /v1/reddit/search for the posts, GET /v1/reddit/posts/comments for the thread, GET /v1/twitter/profiles/{handle} for the author, GET /v1/tiktok/videos/transcript and GET /v1/youtube/videos/transcript for video. For developer communities there's GET /v1/hackernews/search and GET /v1/github/repositories.

Coverage differs too. Octolens covers Reddit, X, LinkedIn, Hacker News, GitHub, YouTube, TikTok, Bluesky, DEV and Stack Overflow, and adds podcasts, news, web, newsletters and reviews on Scale. It doesn't cover Instagram, Facebook or Threads. Social Fetch has GET /v1/instagram/search/hashtags, GET /v1/threads/search and Instagram and Facebook post and comment endpoints.

Octolens allows 500 API requests an hour per organisation, across all keys, and returns 429 with a Retry-After header past that. Exports have their own limits: up to 50,000 mentions per export, 5 exports an hour and 2 at a time. Its on-demand search, POST /api/v2/search, covers the last 1, 7 or 30 days, and new results count against your monthly mention quota. Past the quota, mentions keep flowing and bill at the plan rate ($0.013 on Pro, $0.01 on Scale) unless you set a spend cap or switch flex mentions off.

Social Fetch has no request quotas on paid metered routes and no enforced RPS cap; credits are the only limit. For reliability we recommend staying under ~500 concurrent requests. A search costs 1 credit whether it returns two posts or twenty, so a keyword that suddenly trends doesn't change the bill.

Octolens sends webhook, Slack and email notifications from a feed of mentions on every plan. Social Fetch Monitors do the same job one level lower: you pick a source, its params and a schedule with POST /v1/monitors, and we poll, diff against what we've seen, and POST only new items to your endpoint.

Watchable sources are X profile tweets and search; Reddit subreddit posts, subreddit search and site-wide search; TikTok profile videos, video search and top search; and Instagram posts, reels, tagged posts and hashtags. Anything else is on-demand search you schedule yourself. Checks can run as often as every minute, or every 5 minutes on X. Each delivery is signed with an HMAC in the socialfetch-signature header and retried on failure, and the event id stays the same across retries so you can dedupe.

Creating and editing monitors is free. Each check bills the route's normal price, 1 credit for the four search sources, and ?dryRun=true returns the monthly cost before you commit. A check returns one page of results, so a busy keyword can need a shorter interval to avoid missing posts. Details are in the Monitors docs.

Octolens's AI relevance scoring is its strongest feature. It uses your company context to label each mention relevant or not, so a keyword like "Apple" or a common word doesn't flood the feed. If you want that done for you, Octolens does it well.

Social Fetch doesn't ship a relevance model. You run the filter with the model you already use, which means you write and own the prompt. In code, send each post's text and a short description of your product to Claude or GPT and keep only what it marks relevant. With an agent, connect the MCP server and ask it to search and filter in one step. Either way, treat post text as untrusted input, since anyone on the internet wrote it.

The standard job: track 10 keywords on X, Reddit, TikTok and Instagram, checked hourly. That's 10 × 4 × 24 × 30 = 28,800 checks a month. On Social Fetch the four search sources (GET /v1/twitter/search, GET /v1/reddit/search, GET /v1/tiktok/search and GET /v1/instagram/search/hashtags) cost 1 credit per check, and 40 monitors fit the Growth plan. Octolens prices by keyword and by mention instead.

Octolens is a strong mention feed for B2B and devtools teams: AI relevance scoring, Slack routing and developer sources like Stack Overflow and DEV, priced by keyword and mention. Social Fetch is the data API: live keyword search on X, Reddit, TikTok, Instagram and more, Monitors with signed webhooks, and profiles, comments and transcripts behind the same key, billed per request with no enforced rate cap. For the standard 10-keyword job it costs $79/mo against Octolens Pro's $159/mo, and it covers Instagram. Pick Octolens if you want relevance filtering done for you; pick Social Fetch if you want the raw data and your own model.