# FMInference/FlexLLMGen

Running large language models on a single GPU for throughput-oriented scenarios.

Repository: https://github.com/FMInference/FlexLLMGen
Canonical: https://ross.abutalabs.com/products/flexllmgen
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, gpt-3, high-throughput, large-language-models, machine-learning, offloading, opt
Archived: true
Last push: 2024-10-28T03:05:41+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 92
- inputs: {"age_days": 1295, "days_push": 674, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 9352, forks 591 (observed 2026-08-28T04:10:30.789390+00:00)

## What it is
FlexLLMGen is a high-throughput generation engine for running large language models on a single GPU with limited memory. It uses IO-efficient offloading, compression, and large effective batch sizes to maximize tokens-per-second for throughput-oriented batch inference workloads.

## Use cases
- run large language model inference on a single consumer GPU
- classify or extract information from millions of documents in batches
- benchmark LLMs like HELM on limited hardware
- process a company's private document corpus with an LLM overnight
- run OPT or GPT-style models with limited GPU memory
- reduce inference costs by trading latency for throughput

## When to choose
- you need to process large batches of text and throughput matters more than latency
- you only have a single commodity GPU but want to run large models
- you are running offline batch jobs like data wrangling or benchmarking

## When to avoid
- you need low-latency interactive chat responses
- you have enough powerful GPUs to hold the whole model in memory
- you need small-batch or real-time inference

## Facets
- artifact type: library
- maturity: maintenance
- function: llm-inference, machine-learning, gpu-computing
- domain: large-language-models, deep-learning, machine-learning
- platform: python
- tags: offloading, high-throughput, single-gpu, model-compression, batch-inference, gpu, linux

## Member repositories
- FMInference/FlexLLMGen (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:30.789390+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:22:26.739182+00:00, confidence not recorded.
  - readme: https://github.com/FMInference/FlexLLMGen (fetched 2026-08-28T04:10:30.789390+00:00, sha 866f38aaf8b5)
- Data as of 2026-08-30T08:39:29.467469+00:00.