All tags
Person: "rseroter"
not much happened today
ornith-1.5 qwen3.8-27b claude-opus-5 kimi-k3 glm-5.2 grok-4.5 gpt-5.6-luna grok-4.6 glm-5.3 trueforge ornith vllm ollama unsloth qwen arena valsai deepseek truefoundry claude model-compression quantization reinforcement-learning agent-evaluation plugin-architecture open-agent-runtime cost-efficiency session-management tooling benchmarking ornith_ unslothai danielhanchen arena valsai zhihufrontier theturingpost truefoundry omarsar0 kimmonismus bradenjhancock dbreunig rseroter claudedevs
Ornith-1.5 launches as a new open-weight model family with 9B dense, 35B MoE, and 397B MoE variants under MIT license, featuring quantized formats like FP8, GGUF, MLX, and NVFP4 and showcasing end-to-end self-improvement capabilities. Compression techniques improve accuracy and efficiency, with Qwen3.8-27B GGUFs using Dynamic V3 achieving 10% higher accuracy and 1-bit quantization retaining 77% BF16 accuracy on 8GB RAM. Agent evaluation boards highlight models like Claude Opus 5 (High), Kimi K3, GLM 5.2, Grok 4.5, and GPT-5.6 Luna leading in quality and value. DeepSeek Harness (DSH) introduces a plugin-based open agent runtime architecture optimized for extensibility and tooling. TrueFoundry open-sources TrueForge, a self-hostable, vendor-neutral agent harness that reduces token usage by 30% and cuts costs by 75% while maintaining accuracy, emphasizing the growing importance of session, environment, memory, and tools layers in agent platforms.