deepseek-v4.1-flash-1017 fixing a real bug in Claude Code and OpenCode, and what each run cost
DeepSeek's October 17 snapshot of V4.1 Flash is the model behind the deepseek-flash name on DeepSeek's API, and it is now callable under its pinned ID deepseek-v4.1-flash-1017 on XHuoAPI. Pinning the snapshot matters for agents: the behaviour you tested is the behaviour you keep. Disclosure: I run XHuoAPI. DeepSeek's own prices are listed next to ours, with the source, so you can compare. The…
DeepSeek's V4.1 Flash model is the backbone of its API, now accessible via the pinned ID deepseek-v4.1-flash-1017 on XHuoAPI. Pinning ensures consistent behavior for agents. DeepSeek prices are listed alongside XHuoAPI's, allowing for cost comparison. The objective was to fix a failing test: orders of $50 or more should have no shipping fee, but total 50 charged a $50 order.
Each tool received the same prompt to identify and correct the bug in shipping.js, with a run only counted if node --test was successful afterward. Claude Code (Anthropic Messages format) version 2.1.310 and OpenCode 2 (OpenAI Chat Completions format) version 2.1.3 both passed the test, each requiring around 20-49 seconds per run.
Claude Code processed approximately 58,000-71,000 input tokens per run, generating 550-750 output tokens, costing about $0.02 per run at XHuoAPI's Standard price (checked on 2026-10-26). DeepSeek charges peak hours rates at twice the off-peak rate, while XHuoAPI Standard, Saver, and Spot tiers offer reduced costs with varying trade-offs in stability.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
