[k__]

My 2 weeks with DeepSeek V4:

Pro is ~50% more expensive than Flash.

Both need babysitting.

Plan, split in small tasks, give it docs, types, tests, linter, best practice examples, etc.

Always start a new session when starting a task.

Do regular manual sanity checks, and tell it to find issues in the codebase.

I pay like $1,50 per day for Pro.

[p1necone]

I use GLM-5.2 as an orchestrator model which delegates to Deepseek subagents (v4 flash or pro depending on complexity) and it works pretty well for a quite complex compiler codebase when I do deep enough up front planning for features/fixes.

If you believe the benchmarks Deepseek v4 is pretty shitty at long running work in large codebases, but really really good at self contained algorithmic/math reasoning - which is basically ideal for a compiler for a language with a relatively complex type system. And with its cache pricing it's very cheap.

GLM-5.2 is too expensive at api pricing for my taste though even when it's not producing the bulk of the output tokens - I pay for the mid-tier subscription and switch the orchestrator over to other models via open router when I run out - Minimax M3 feels okay but definitely a step down from GLM-5.2.

[h2aichat]

I had a similar experience

[5701652400]

very simlar experience.

I would also add that I run it this way ~12hour a day non-stop. 300M / tokens per day (99.7% cache hit).