Drei Coding Agents im Vergleich: ein lokales Modell auf 6GB GPU
In einem Test lief dasselbe lokale Modell auf einem RTX 3060 Laptop mit 6GB VRAM in drei Coding Agents. Während pi und Codex alle zwölf Aufgaben lösten, brachte Claude Code es nur auf sieben. Dabei etwa behauptete Claude Code, eine Seite sei fertig, obwohl der Ordner leer blieb.
Posted on Oct 16 • Originally published at kunbuilds.com • AI-assisted
Claude Code told me the web page was done. The folder was empty.
That was the worst moment of a small test: one local model on a 6GB laptop GPU, plugged into three coding agents. Same model, same tasks. Only the agent changed.
Every request starts with the agent's system prompt and tool definitions. First prompt eval in the llama-server log:
With a 32K context, Claude Code fills about half of it before doing anything, and reading that prompt takes about 8.6× longer than pi's.
12/12 on both models, zero tool errors in the logs, fastest bug fix (47 s median). I also trimmed its prompt to about 750 tokens with only read/edit/bash enabled. Still 12/12, same speed. Harder tasks might separate the two; these didn't.
12/12, but LFM made 12 tool-argument errors, mostly integers sent as strings:
failed to parse function arguments: invalid type: string "96787", expected i32 Enter fullscreen mode Exit fullscreen mode It retries and recovers. The cost is time: 85 s median for the bug fix, 599 s on the first run.
Prompt is too long · automatic compaction failed: summarization produced empty response Enter fullscreen mode Exit fullscreen mode Timer page: 3/3, but an 881 s median, and one run hit the 900 s cap. Qwen: 0/3. For the page and the pelican it replied that the task was complete. The working folder had no files. An error tells you something failed. A false "done" only shows up when you check the folder yourself.
Called directly, LFM's pelican floats above the bike; inside an agent it actually sits on it. Qwen's drawings are more complete. Claude Code + Qwen produced no SVG.
Full data and configs: https://kunbuilds.com/en/posts/agents-local-model-6gb/