AI tools
LLM runtimes
5 listings are ranked by release freshness and adoption until verified reviews arrive; then review quality, volume and vendor responsiveness join the score. Open any row to see its receipt. See the formula
- 4 fresh released in the last 90 days
- 0 aging no release for 90 to 179 days
- 1 stale no release for 180 days or more
-
Ollama
Runtime to download and run large language models locally, with a CLI and a REST API.
Updated 62 days ago1 93.5 =Show how this rank is calculated
Rank receipt: Ollama#5 overall · #1 in category93.5/100- Platform readiness +40.0 100.0 × 40%
- Release freshness +31.6 90.4 × 35%
- Adoption signals +21.9 87.6 × 25%
-
llama.cpp
C/C++ inference engine for LLMs on CPUs and GPUs, using the GGUF format and quantization.
Updated 61 days ago2 93.0 =Show how this rank is calculated
Rank receipt: llama.cpp#6 overall · #2 in category93.0/100- Platform readiness +40.0 100.0 × 40%
- Release freshness +31.7 90.7 × 35%
- Adoption signals +21.3 85.2 × 25%
-
LocalAI
Self-hosted, OpenAI-compatible API that runs text, image and audio models on local hardware.
Updated 62 days ago3 85.2 =Show how this rank is calculated
Rank receipt: LocalAI#64 overall · #3 in category85.2/100- Platform readiness +34.0 85.0 × 40%
- Release freshness +31.6 90.4 × 35%
- Adoption signals +19.6 78.2 × 25%
-
vLLM
High-throughput LLM serving engine with paged attention and an OpenAI-compatible server.
Updated 62 days ago4 80.3 =Show how this rank is calculated
Rank receipt: vLLM#92 overall · #4 in category80.3/100- Platform readiness +28.0 70.0 × 40%
- Release freshness +31.6 90.4 × 35%
- Adoption signals +20.7 82.7 × 25%
-
LM Studio
Discover, download and run local LLMs offline on Mac, Windows and Linux with an intuitive desktop interface and local OpenAI-like server.
Stale: 646 days without a release5 40.0 =Show how this rank is calculated
Rank receipt: LM Studio#367 overall · #5 in category40.0/100- Platform readiness +40.0 100.0 × 40%
- Release freshness +0.0 0.0 × 35%
- Adoption signals +0.0 0.0 × 25%
Data from Advisor (AI agent) Last sync Computed on Sponsorship weight 0%, never counted Read the full formula