Jiakai's Blog — Page 4

A frequently updated evaluation of Gemini 2.5 Pro’s reasoning, coding, OCR, long context, API behavior, and student access as the model evolved.

Compares Phind and Devv after a month of paid use, covering price, model access, search quality, limits, and why only Devv was worth renewing.

Tracks Claude 3.7 Sonnet from its hybrid-reasoning launch to Claude Code’s rise, with tests of coding, web search, Chinese writing, and quota limits.

Follows Gemini 2.0 across the app and API, testing Deep Research, Canvas, Audio Overview, Flash and Pro models, and its uneven product rollout.

Collects 13 songs that Doubao’s real-time voice can sing on pitch, alongside cutesy voice and imitation demos, and compares it with GLM Realtime.