Evaluate long deliveries separately from long inputs
A four-part task plan and runnable synthetic ledger keep missing runs, retries, truncation and human handoffs in evaluation records.
Read articleFrom the IteraCompute team
Blog
Engineering notes, practical guides and product updates from the IteraCompute team.
Track agent request costs and validate JSON before accepting results
Two offline Python examples show how to preserve unknown tool charges and reject JSON records that parse correctly but disagree with their source.
Read article
Preserve tool-call history across a JSON save and restore
A runnable Python example checks what survives when tool-call history is saved to JSON and restored, including assistant fields and tool result IDs.
Read article
API key concurrency: model one account-wide admission budget
A runnable offline example shows how two API keys can exceed one shared account concurrency budget when each key admits requests independently.
Read article
LLM retry costs: keep unknown usage in an attempt ledger
An answer can pass your application's acceptance check while its cost record remains incomplete. This matters when a paid coding tool retries a request: keeping only
Read article
DeepSeek V4 Pro API price cut: over 54% below official peak rates
Our listed Pro 0813 rates are now $0.60 input and $1.79 output per million tokens. See the reductions and compare official peak-period costs.
Read article
DeepSeek V4.1 Flash API prices cut again: compare the costs
We've lowered Flash input and output prices again. Compare the new rates and total request costs with DeepSeek's official peak and off-peak pricing.
Read article