RWKV-6 7B
Attention-free RNN-based 7B LLM with linear-time inference and strong multilingual coverage.
Retired
RWKV-6 7B appears to be discontinued or shut down — end-of-life; superseded by RWKV-7 This page is kept as an archive record and is excluded from search engines.
About
RWKV-6, codenamed Finch, is the sixth generation of the RWKV architecture, an attention-free sequence model that combines transformer-style training with RNN-style inference. The 7B World model was trained on trillions of multilingual tokens and released under the Apache 2.0 license, making it one of the most capable non-transformer open models of its size in 2024.
Because it uses linear recurrences instead of self-attention, inference cost stays constant per token regardless of context length, which makes it attractive for long-context and edge deployment. Note that the RWKV team has since marked the RWKV-6 series as end-of-life in favor of RWKV-7, so the 7B Finch checkpoints now serve mainly as archives and research artifacts.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 129; Trending score 74/100 from researcher assessment.