Starling-LM-7B
Open 7B chat model tuned with reinforcement learning from AI feedback.
About
Starling-LM-7B is an open chat model from UC Berkeley researchers and Nexusflow, fine-tuned from OpenChat 3.5 with RLAIF — reinforcement learning from AI feedback using the GPT-4-labeled Nectar ranking dataset. The beta release scored 8.12 on MT-Bench with GPT-4 as judge, strong for a 7B model of its era.
It is a 2023/2024-era model, so it lags far behind modern small models on knowledge and reasoning, and its license carries an unusual condition against competing with OpenAI. Still useful as a lightweight, fully local conversational model.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
MT-Bench 8.12 (beta, GPT-4 judged, team-reported)
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 Evidence: 2023/2024-era model; superseded by newer small models; Exact beta release day not pinned — releasedAt left null (paper Nov 2023, beta Mar 2024); Manual review: historically important RLAIF work; license has unusual non-compete-with-OpenAI condition. License: Apache-2.0 with a condition against use competing with OpenAI (per Ollama README); alpha was non-commercial.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · 2023/2024-era model; superseded by newer small models
- · Exact beta release day not pinned — releasedAt left null (paper Nov 2023, beta Mar 2024)
- · Manual review: historically important RLAIF work; license has unusual non-compete-with-OpenAI condition