avifenesh/memra

Inference Engines

Rust + CUDA inference engine for NVIDIA RTX PRO 6000 Blackwell and RTX 5090. Serves safetensors and GGUF over an OpenAI-compatible API, with per-device tuned defaults and speculative decode gated byte-identical to plain decode. Hosted instance: inference.tiyuvta.ai

GitHub →Website →

4.1

GitHub Metrics

Stars

Forks

Open Issues

Watchers

Contributors

Weekly Commits

Language

OpenEdge ABL

License

MIT

Last Commit

Oct 6, 2026

Created

Sep 1, 2026

Latest Release

v0.138.0

Release Date

Sep 11, 2026

Synced: Oct 27, 2026

Quality Scores

Documentation Qualityw: 20%

4.4

Has docs site (https://inference.tiyuvta.ai). Description: 265 chars. Stars signal: 4. Contributors: 5. Score: 4.4/10

Community Healthw: 20%

0.6

Stars: 4. Contributors: 5. Watchers: 1. Forks: 1. Score: 0.6/10

Maintenance Velocityw: 15%

7.2

Last commit: 20d ago. Weekly commits: 0. Latest release: v0.138.0. Score: 7.2/10

API Design & DXw: 20%

5.4

Has documentation site. Permissive license: MIT. Popularity signal: 4 stars. Score: 5.4/10

Production Readinessw: 15%

3.5

Battle-tested: 4 stars. Peer review: 5 contributors. Versioned: v0.138.0. Licensed: MIT. Age: 0.2 years. Maintenance: last commit 20d ago. Score: 3.5/10

Ecosystem Integrationw: 10%

3.8

Fork interest: 1. Ecosystem: OpenEdge ABL. Integration-friendly: MIT. Adoption: 4 stars. Has web presence. Score: 3.8/10