🟧 Hacker News 2d ago
Gemma 4 Release Triggers Debate About Tool Calling Implementation Issues
Gemma 4 release exposed systemic reliability issues where local model runners (Ollama, LM Studio) rushed launch-day support with broken tokenizer implementations and failed tool calls. Discussion highlighted trade-offs between inference tools, with performance benchmarks showing Ollama 25% faster than LM Studio on Mac, but recurring pattern of premature releases creating production issues.