Skip to main content
These entries record LaoZhang API announcements from August 2025, newest first. Prices and availability describe the state at the time; current models and prices are in the model and pricing catalog.

​
August 2025

​
August 26, 2025: DeepSeek V3.1 launched

  • New model: deepseek-v3-1-250821, with a thinking mode for step-by-step reasoning and a non-thinking mode for fast responses.
  • Prices at launch: $0.50 per 1M prompt tokens and $1.50 per 1M completion tokens.
  • 128K context window, Anthropic API format support, and beta function calling.
  • Mode names: deepseek-chat for non-thinking mode and deepseek-reasoner for thinking mode.

​
August 25, 2025: Kimi K2 release version

  • New model: kimi-k2-250711, the Volcano Engine release of Kimi K2. It replaced the preview model kimi-k2-0711-preview.
  • Actual charges followed the console price at the time.
  • Existing preview users were advised to switch to the release version.

​
August 8, 2025: GPT-5 series launched

  • New models:
    • gpt-5
    • gpt-5-2025-08-07
    • gpt-5-chat-latest
    • gpt-5-mini / gpt-5-mini-2025-08-07
    • gpt-5-nano / gpt-5-nano-2025-08-07
  • Priced the same as OpenAI’s official prices at the time: $0.05 to $1.25 per 1M prompt tokens and $0.40 to $10.00 per 1M completion tokens, depending on the model.
  • Request notes:
    • The models supported the /v1/responses endpoint.
    • temperature had to be 1 or omitted.
    • max_completion_tokens replaced max_tokens.
    • gpt-5-chat-latest used /v1/chat/completions.

​
August 7, 2025: Claude Opus 4.1 launched

  • New model: claude-opus-4-1-20250805, an update to Opus 4 focused on coding and reasoning, with a 200K context window.
  • Priced the same as Anthropic’s official prices at the time, with usage-based billing.

​
August 6, 2025: OpenAI open-weight models launched

  • New models: gpt-oss-120b (117 billion parameters) and gpt-oss-20b (21 billion parameters).
  • 128K context window, low, medium, and high reasoning levels, a mixture-of-experts architecture, and the Apache 2.0 license.
  • OpenAI stated that gpt-oss-120b runs on a single 80 GB GPU and gpt-oss-20b on devices with 16 GB of memory.
Back to the June–August 2025 archive index