Felo AI

Felo AI is a multilingual AI search and creation workspace built around live-web retrieval, source-backed synthesis and model choice. The current product goes beyond a single search model: Felo exposes multiple frontier models, agentic research with live web context, document and file analysis, and output tools for presentations, documents, images, mind maps and web pages. Its strongest differentiator is cross-language retrieval: users can ask in one language and synthesize sources found in others. ¹²

The platform now extends into developer workflows. Felo's API platform shares the same credit system as the consumer product and offers model access plus ready-made Search, X Search, Web Fetch, PPT, Image, Mindmap and Page APIs. Agent-client support explicitly includes Claude Code, Codex and OpenClaw, so the product can act as both an end-user research surface and a programmable research/content backend. ³

Felo is moving quickly, which is both an advantage and a source of volatility. Recent 2026 product pages expose new model launches and agent capabilities soon after release. That makes model choice and current-web research attractive, but it also means feature availability, quotas and pricing can change faster than in a more static search product. ¹³

Felo's current API pricing page documents a unified credit system shared across Felo products. Standard accounts are free and receive 200 credits daily. The paid subscription is $14.99 per month or $149.99 per year and includes 15,000 monthly credits plus the daily 200-credit grant, with broader model and tool access. Felo defines 1,000 credits as $1 of credit value; Enterprise uses custom allocations. ³

A current Search FAQ separately describes the $14.99 monthly Pro subscription and its professional-search allowance. Because Felo has been migrating functionality into the shared credit system, the credit balance is the more durable way to understand current usage economics than older per-search descriptions alone. Heavy model, search, slide or image use can draw from the same pool. ⁴³

Independent exact-product coverage is still limited. A May 2026 hands-on teardown highlights Felo's multilingual retrieval and research-to-output workflow as the real product wedge, while also noting that performance claims are strongest in East Asian plus English use cases and that enterprise conversion and broader-language quality remain less transparent. This is useful current qualitative evidence, not a controlled benchmark. ⁵

The fixed panel provides stronger evidence for the mobile surface than for the web product. The Japanese App Store currently shows Felo Search at 3.7/5 from 83 ratings. Reviews praise multilingual retrieval, source breadth, summaries and mind maps, but some complain about mobile/web feature gaps and file-upload limitations. Google Play shows 3.9/5 from 587 reviews and 100K+ downloads in the current snapshot. Both are exact-product signals, but they are weighted as partial-surface evidence because the broader Felo web/API workspace is materially richer than the mobile apps. ⁶⁷

G2 has an exact Felo profile describing the multilingual search/workspace product, but it currently has zero reviews. Its displayed 0.0 is therefore an empty-state placeholder, not a negative rating and not a rating input. ⁸

Privacy documentation is less easy to evaluate on the consumer surface than pricing and product capabilities. Felo's file-upload documentation says uploaded files are used only to answer the user's task, with a seven-day retention period for standard use but permanent retention for Pro files. The Enterprise privacy policy says user content is not used for model training without explicit consent, although it separately permits use of other personal information for AI development/improvement purposes under that enterprise policy. Buyers handling sensitive data should verify which policy and retention controls apply to their exact plan. ²⁹

Felo is best for multilingual knowledge workers, analysts, researchers and developers who routinely need to search across markets and languages, compare model outputs and turn research directly into documents, presentations or other artifacts. Its API and agent-client support also make it interesting for teams that want one credit pool across interactive and programmatic AI workflows.

It is less compelling for users who only need a stable English-first answer engine, teams that require extensive independent benchmark evidence before adoption, or organizations that need very explicit consumer-plan data-retention and governance terms without further procurement review.