AI Engineering colibrì: How a Single C File Runs a 744B-Parameter MoE Model on 25 GB of RAM
Discover how colibrì runs the 744B-parameter GLM-5.2 model on consumer hardware by streaming experts from…
Category: AI Engineering
AI Engineering 
AI Engineering Google’s TorchTPU Runs PyTorch Natively on TPUs at 2.01 Cents per Million Tokens, but That Is a Hardware Cost Benchmark, Not Your API Bill
Google and Meta open-sourced TorchTPU, a native PyTorch backend for TPUs. The benchmark claims are…
AI Engineering llm-d 1.0 Runs 3,000 Coding Agents at 5 to 10x Lower Token Cost, but Every Benchmark Is a Vendor Demo
llm-d 1.0 adds multi-tenant QoS, zero-downtime upgrades and agent-aware routing. IBM claims 5 to 10x…
AI Engineering DeepSeek Open-Sources DeepInfer, the Engine Serving 155 Trillion Tokens a Day, but It Leaves Out the KV Cache and Load Balancers and Runs Best on 128 GPUs
DeepSeek open-sourced DeepInfer, the engine it says serves 155 trillion tokens a day at $38…
AI Engineering OpenAI Says GPT-6.1 Sol Went From 5.4% to 86.6% on COBOL Migration, but That Test Isn’t Public and Its Open Benchmark Grades Just 272 Tests
OpenAI says GPT-6.1 Sol passed 86.6% of its COBOL migration tasks, up from 5.4%, and…
AI Engineering Google’s Open-Source Harness2 Lifts GLM-5 Past Gemini 3.5 Flash Without Touching Its Weights, but Every Task Runs at Least Four Times and AI Models Do the Grading
Google Research open-sourced Harness2, which edits an agent's prompts, tools, memory and skills during a…
More articles are loading as you scroll.