DeepSeek
DeepSeek's current V4 API line, covering V4-Pro and V4-Flash with 1M context, thinking mode and tool calls.
Quick answer
V4 is the current API line, with separate migration and pricing implications from the broader DeepSeek brand.
The English DeepSeek docs list deepseek-v4-pro and deepseek-v4-flash as the stable public IDs, currently routing to V4-Pro-0813 and V4-Flash-0731. Both support thinking and non-thinking modes, JSON output, tool calls and chat prefix completion; FIM completion is available in non-thinking mode. Billing now varies by peak and off-peak UTC windows while model IDs remain unchanged.
Topic guideA source-backed map of DeepSeek V4 models, API pricing, native Harness, third-party agent runtimes and practical integration guides.
Editorial verdict
Developers migrating DeepSeek integrations to current V4 model names and long-context reasoning workflows.
Avoid starting new projects on deepseek-chat or deepseek-reasoner aliases.
V4 is the current API line, with separate migration and pricing implications from the broader DeepSeek brand.
Off-peak V4-Flash: $0.007/$0.22/$0.66; V4-Pro: $0.022/$0.66/$1.98 per M cache-hit/cache-miss/output tokens; peak rates are double
Topped-up balance, Granted balance, Platform billing
Commercial use should follow the current product, API, model license and billing terms.
Review prompt, file, media upload, retention and training-use terms before sensitive workloads.
Use deepseek-v4-pro for agent/coding tasks and deepseek-v4-flash for faster or lower-cost workloads.
Use the documented thinking toggle and effort controls for reasoning-heavy requests.
Model names, quotas, release status, regional access and commercial terms can change quickly; recheck official sources before procurement or production use.
qwen
kimi-k2-api
zhipu-glm
Pricing · EN · verified 2026-08-17
Lists the current V4-Pro-0813/V4-Flash-0731 routes, context and output limits, features, concurrency and active peak/off-peak prices.
Documentation · EN · verified 2026-05-17
Documents thinking toggle, effort control and reasoning_content behavior.
Documentation · EN · verified 2026-05-17
Confirms the 2026-04-24 V4 release and 2026-07-24 alias deprecation.
Last checked: 2026-09-07
No approved reviews yet.