PulseAugur
EN
LIVE 03:02:45

Four large local LLMs benchmarked on RTX 5090 with 128GB RAM

By PulseAugur Editorial · [1 sources] ·

A technical benchmark compares four large local language models (LLMs) running on a single RTX 5090 GPU with 128 GB of RAM. The models tested include DeepSeek-V4.1-Flash (748B parameters), GLM-5.3-Flash (321B parameters), Qwen3.8 Flash-Next, and Ornithogalum, using different runtimes like Project Maya, Strata, and ExLlamaV3. The benchmark measured decode speed, prefill performance with long contexts up to 125K tokens, drafting capabilities, and quality on practical tasks. Results indicate varying strengths for each model and runtime configuration, with Qwen3.8-Flash-Next achieving high token generation speeds. AI

IMPACT Provides practical insights into local LLM performance and hardware requirements for users.

RANK_REASON Technical benchmark comparing performance of multiple LLMs on specific hardware. [lever_c_demoted from research: ic=1 ai=1.0]

Read on dev.to — LLM tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Four large local LLMs benchmarked on RTX 5090 with 128GB RAM

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Technical benchmark comparing performance of multiple LLMs on specific hardware. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
infra, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
2 days old
Coverage has settled into its steady-state source set.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Seppe Gadeyne ·

    Four big local LLMs on one RTX 5090 and 128 GB of RAM, benchmarked side by side

    <p><strong>One RTX 5090, 128 GB of RAM, and four open models from 125B to 748B parameters.</strong> Over the past weeks I got each of them running on the same workstation, on three different runtimes. On October 21, 2026 I measured all four on the same day, one after the other, w…

RELATED ENTITIES

RELATED TOPICS