06-22-2026, 01:45 PM
Updating this thread with June 2026 additions:
New entries:
Local model standouts this month: Qwen3 14B and Gemma 3 27B are the ones I keep going back to for day-to-day tasks. Both run well on an RTX 4070.
New entries:
- Claude Sonnet 4.6 - Anthropic's latest mid-tier model. Noticeably better at following complex multi-step instructions than previous versions. Good coding assistant.
- Gemini 2.5 Pro - Google's strongest model as of mid-2026. Long context window (1M tokens), genuinely useful for large codebase analysis.
- Qwen3 235B (MoE) - Alibaba's open-weight mixture-of-experts model. Competitive with GPT-4 class on benchmarks, runs surprisingly well on high-end consumer setups.
- Mistral Large 2 - Still a strong choice for European data residency requirements (hosted in EU by default).
Local model standouts this month: Qwen3 14B and Gemma 3 27B are the ones I keep going back to for day-to-day tasks. Both run well on an RTX 4070.