Feel free to contribute.
gemma4:e4b (9.6GB) - Ollama
Practically useless for complex planning tasks it fails expanding the code base and working alone one tasks
gemma-4-26b-a4b - LM Studio
It is working much better - even if it is the same model; but much larger - expanding code and planing properly.
LM Studio is unstable v0.4.10
gemma4:26b / gemma4:26b-a4b-it-q4_K_M - Ollama
As good as gemma-4-26b-a4b - LM Studio, but faster and stable.
Qwen 3.5 - LM Studio
Currently not working properly due to an LM Studio Bug
https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1592
Benchmarks
Zähle mir die letzten 5 Bundeskanzler der Bundesrepublik Deutschland auf und nenne zudem kurz in einer Tabelle, wie lange diese regiert haben und ihre am meisten gefeierte Leistung während ihrer Regierungszeit.
AMD 7900 XT 20 GB on Windows
| Model | Provider | Tokens | Speed | Time | Agent Coding | Status |
|---|---|---|---|---|---|---|
| gemma-4:26b-a4b-it-q4_K_M | Ollama | 1064 | 29.25 tok/s | 36.38s | ❌ Not usable | ✅ Stable |
| gemma-4-26b-a4b | LM Studio | 2460 | 50.72 tok/s | 48.5s | ❌ Not usable | ✅ Stable |
| gemma-4-26b-a4b-it-claude-opus-distill | LM Studio | 841 | 71.37 tok/s | 11.78s | ❌ Defective (tools not working) | ❌ Defective |
| qwen3.6-35b-a3b | LM Studio | 3974 | 33.56 tok/s | 118.4s | ✅ | ✅ Stable |
| qwen3.6-35b-a3b | Ollama | — | — | — | ❌ Timeout | ❌ Timeout (>4 min) |
| glm-4.7-flash-opus-4.5 | LM Studio | — | — | — | ❌ Deadlock | ❌ Instable (deadlock) |
Recommendations
- Qwen 3.6 via LM Studio is the recommended choice for complex planning tasks. It is stable and handles code expansion properly, despite the slower generation time (~118s).