Guides
Short, source-linked how-tos for API model names, off-peak pricing, LLM API price comparisons, Claude Code, Search/Fetch, error codes, and installing DeepSeek Harness (dsh). Independent resource, not affiliated with DeepSeek.
How these guides relate to the official docs
DeepSeek's primary documentation lives at api-docs.deepseek.com. This directory does not mirror every page. Instead, each guide here answers one recurring integration question — model strings, off-peak billing windows, cross-vendor list prices, Claude Code wiring, Search/Fetch billing, HTTP errors, or installing @deepseek-ai/dsh — and links back to the exact official section we checked.
When the API change log moves faster than a long-form article, we prefer dated tables and short FAQs over prose that ages quickly. The model name migration guide tracks rename announcements; the API pricing directory holds the live DeepSeek token table; the news timeline records dated releases. If a figure on this site disagrees with DeepSeek's pricing page, trust the pricing page.
For agent workflows, start with Claude Code or the dsh command reference, then read error codes before you ship retries. For budgeting across vendors, use the cost-effective LLM comparison alongside your own evals — list price per million tokens is only one input.
DeepSeek API model names: what to use now
Current deepseek-flash and deepseek-v4-pro strings, retired aliases, and migration steps.
DeepSeek off-peak pricing: hours and how much you save (Oct 2026)
Peak UTC windows, ET/PT/UK/CET/IST/SGT tables, and half-price off-peak billing from the official pricing page.
Cost-effective AI models: LLM API pricing comparison
Sourced per-million-token API prices across major providers and hosted open models, with use-case notes.
Use DeepSeek models in Claude Code
Anthropic-compatible base URL, env/settings for Claude Code, Codex, and OpenCode pointers.
DeepSeek Search and Fetch API
Search and Fetch endpoints, billing rules, Harness/Codex billing start date, and rate limits.
DeepSeek API errors and rate limits
Official HTTP error codes, concurrency limits, and when to retry.
DeepSeek Harness CLI (dsh)
Official npx install, npm channels, and source checkout commands.