InternLM2
Open Chinese LLM family (1.8B/7B/20B) with strong long-context and RLHF alignment work.
About
InternLM2 is the second-generation open language model from Shanghai AI Laboratory's InternLM team, released on January 17, 2024 in 7B and 20B sizes (a 1.8B variant followed on January 31). The technical report (arXiv:2403.17297) documents a full pipeline: pretraining on text, code, and long-context data, supervised fine-tuning, and a novel Conditional Online RLHF strategy designed to reconcile conflicting human preferences.
The family is notable for its long-context story: trained natively on 32K tokens with extrapolation to 200K in chat variants, and near-perfect needle-in-a-haystack performance at 200K. Code is Apache 2.0 and weights are open for academic research with free commercial use available after a license application — a gated-commercial model rather than fully OSI-open, which users should note before shipping products on it.
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 — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 80; Trending score 62/100 from researcher assessment.