Technology ❯ Artificial Intelligence ❯ Model Evaluation

Performance Metrics

Benchmarking User Feedback Benchmark Testing Comparative Analysis User Experience Reasoning Capabilities Cost Efficiency Accuracy Benchmark Scores User Satisfaction

OpenAI Cuts GPT-5.6 Luna Prices 80% and Terra Prices About 20%

OpenAI says the reductions will expand enterprise access by lowering per‑token costs.

Anthropic Adds Opus and Sonnet to Claude's Voice Mode

Kimi Releases Open 2.8 Trillion‑Parameter Model K3 and Sets Weight Release for July 27

OpenAI Builds GPT-Red to Hunt Prompt‑Injection Flaws

Thinking Machines Releases Inkling, an Open-Weights Multimodal MoE Model

Chinese AI Firms Slash Training and Inference Costs With Sparse MoE and H800 Workarounds

OpenAI Tests GPT‑5.6 and Pro Variant Ahead of Possible Late‑June Launch

Chinese Team Opens 1.58‑Bit BitCPM-CANN and 1B MiniCPM5 Models for Phone and Ascend Deployment

RAG Practice Coalesces Into a Seven-Step Playbook Focused on Retrieval

Gemma 4 Goes Local With Tool Calling on Phones and Apple Silicon

Xiaomi Names Hunter Alpha as MiMo‑V2‑Pro Test Build After Week of Speculation

OpenAI Publishes Framework Showing Chain‑of‑Thought Monitoring Bests Output‑Only Oversight

Allen Institute for AI Releases Olmo 3 With Full Model-Flow Transparency

OpenAI Tests ChatGPT Group Chats in Four Countries, Powered by GPT-5.1

Anthropic-Led Study Finds 250 Poisoned Files Can Backdoor Large AI Models

Anthropic-Led Study Finds About 250 Poisoned Documents Can Backdoor LLMs Regardless of Size

OpenAI Releases GDPval Work Benchmark as Claude Leads Early Results

DeepSeek Releases Open-Source V3.2‑Exp With Sparse Attention, Cuts API Prices as Huawei Offers 0‑Day Support

Google Releases VaultGemma, an Open 1B-Parameter LLM Trained With Differential Privacy

ReFactX Proposes Retriever‑Free Grounding as Travel Study Benchmarks RAG Strategies