Blog

A practical comparison for agent workflows, coding, data analysis and computer use, with benchmarks, API prices and a plan for your own evaluation.

MiniMax launched M3.1 Flash with API rates of $0.15 / $0.60 per million tokens. A practical evaluation plan for support, CRM and coding workflows.

Aleph Alpha has released Kolibri-1’s open weights. For teams working with German suppliers, evidence handling and unsupported answers matter more than a million-token context window.

Opus 5.5 cuts token and cache costs, but changes thinking, tool selection and conversation handling. Validate the integration first, then measure cost per completed task.

Grok 4.7 builds on 4.6. Migration needs attention to the 200k-token pricing threshold, caching and preservation of encrypted reasoning history.