Prime Intellect: INTELLECT-3

INTELLECT-3 is a text-in, text-out model from Prime Intellect with a 131,072-token context window that matches its maximum completion length, meaning it can both ingest and generate long documents within the same session. It supports tool use and reasoning, which makes it usable for agentic workflows and multi-step problem solving. Structured output support is unconfirmed, so developers who depend on reliable JSON schemas should verify that before committing. On benchmarks, the picture is thin: a blended score of 24.7 drawn from only one independent benchmark leaves its general capability largely unproven by comparison-site standards. Pricing is modest at $0.20 per million input tokens and $1.10 per million output tokens. Buyers who prioritize low cost and a large context window for tool-augmented tasks may find it worth testing, but the limited benchmark coverage means it is difficult to position confidently against better-documented alternatives without running your own evaluations first.