Maintaining local Qwen: turn failures into tests
Eight failures from a home Qwen and RTX 3090 experiment, with controls for upgrades, version changes, handover, testing and operator training.
qwen3.8maintenanceevaluationstraining
Syntalith blog
Start with the problem you need to solve. Explore team training, agents, automation, internal knowledge, custom models and alternatives to costly software.
How DFlash2 speeds up Qwen3.8 and shortens context
In our measurement, DFlash2 raised median decoding from 93.3 to 120.9 tok/s while the context limit fell from 150,000 to 65,536 tokens.
qwen3.8mtpdflash2optimization
One RTX 3090 for a team: measured Qwen concurrency
One card can process requests in sequence or in parallel. We pair our task times with a public 1, 2, 4 and 8-stream RTX 3090 test.
qwen3.8rtx 3090throughputdeployment
One Qwen, four clients: Qwen Code, Codex, Claude Code, OpenCode
Four clients drove the same local Qwen3.8-27B. We compare the tasks, elapsed time, hidden tests and actual code choices. OpenCode won one import; Qwen Code won the long run.
Qwen Code in practice: CSV, pagination and import
Four small tasks revealed more than one large benchmark. Qwen Code read contracts well at medium effort, but hidden tests still found a missing validation rule.
Qwen Code: 26 minutes to a frontend, 48 to acceptance
An empty folder, one sentence and local Qwen on a home RTX 3090. The first ATS dashboard took 26:03. Here are the prompt, output and complete repair cost.
Qwen3.8-27B: 60k, 120k, 150k or 250k context?
Every configuration returned three planted facts, but 250k was almost ten times slower than 60k. Here is how to choose a useful context window.
Qwen3.8-27B quantization: Q4, Q5 or W4A16?
Q4 leaves more room for context, Q5 uses more VRAM, and W4A16 needs a different stack. Here are the profiles that fit and what the test cannot prove.
Qwen3.8-27B on an RTX 3090: results, metrics and limits
A complete record of one home experiment with Qwen3.8-27B: serving profile, timings, tokens, energy, VRAM, coding tests, failures and limits.
Qwen3.8, Bielik or PLLuM? A task-based comparison
Qwen3.8-27B, Bielik-11B-v3.0-Instruct and PLLuM-12B-chat-2512 compared by parameters, architecture, Polish use, licensing, hardware and deployment.
qwen3.8bielikpllummodel comparison
How we isolated the local Qwen agent experiment
Clean fixtures, separate worktrees, hidden checks outside the agent folder, an explicit tool list, telemetry and retained diffs. We also name the gap: shell was not a full sandbox.
local llmqwen3.8securityexperiment
Safe Qwen deployment: threat-model the path from laptop to tool
We decompose the real laptop -> private endpoint -> Qwen -> tool path, showing what the home experiment protected and what a company still needs.
qwen3.8securitydeploymentlocal llm
How to use local Qwen remotely from any device
Qwen can stay on one GPU host while an approved phone, tablet, laptop or office computer reaches it through a private interface. This guide covers three access patterns.
When not to deploy Qwen locally: five results that should stop a project
Five stop signals: an unfinished task, repeated context processing, costly 250k, missed validation and a broken frontend.
qwen3.8local llmapideployment
AI Cost Management in Poland: What to Measure
A provider invoice shows usage. A management view connects that bill to an application, owner and accepted business case. Learn when native reports are enough and when a separate cost layer earns its place.
AI cost managementFinOpsPolandunit economics
AI App, AI Agent or Automation: How to Choose
An app gives people a working interface, automation follows a fixed process and an agent selects steps inside boundaries. Use this buyer router before scoping.
Custom AI App or SaaS: Which Should You Buy?
SaaS usually starts faster. A custom AI app earns a business case when the process is stable, company rules matter and recurring licence costs exceed ownership costs.
Personal AI agent vs ChatGPT vs Copilot: choose
ChatGPT is a general AI workspace, Microsoft 365 Copilot works across Microsoft 365 apps, and a personal agent is a configured process operator.
Bielik or PLLuM in a Company: Hardware and GDPR
A local Polish model brings a hardware bill, license checks and an owner for updates. Compare Bielik and PLLuM with an API before buying a GPU.
PLLuM for business: licence or API?
Check the exact PLLuM model card, licence, data path, infrastructure, and Polish workload before choosing local deployment over an API.