DeepSeek updates V4.1 Flash model and opens search API
要闻
DeepSeek has updated V4.1 Flash to V4.1-Flash-1017 and opened its Search and Fetch interfaces to the public, each priced at CNY 5 per 1,000 calls. The current free arrangement for calls through specified development tools will end at 00:00 Beijing time on November 1, 2026.
DeepSeek has deployed V4.1-Flash-1017 across its official web service, APP and API and opened its Search and Fetch interfaces to the public. Both cost CNY 5 per 1,000 calls, with existing related calls within development tools becoming billable at 00:00 Beijing time on November 1, 2026. According to the company, users accessing the DeepSeek API through DeepSeek Harness, Claude Code or Codex already reach these interfaces through Server-side tool calls, which are currently unbilled. DeepSeek says its search service runs on an internally developed search engine that indexes hundreds of billions of web pages worldwide, supports multilingual search and filters SEO spam, GEO manipulation and low-quality AIGC.
On the model side, DeepSeek says reinforcement learning (RL) based on self-feedback has improved the new version’s ability to build complete projects from scratch in areas such as game development and 3D modeling, while reducing task execution time and output length. The optimizations cover platforms and frameworks including DeepSeek Harness, Tencent CodeBuddy and WorkBuddy. The company recommends upgrading DeepSeek Harness to version 0.2.1-rc.1 or later. Download links for the model weights are available on Hugging Face and ModelScope.
OpenCode discloses security incident affecting some users
要闻
OpenCode disclosed that unauthorized access to a legacy system may have exposed third-party API keys stored in Console, affecting 1.4% of Console users. The company said it had emailed everyone affected and instructed them to revoke and replace their keys and review usage.
OpenCode disclosed unauthorized access to one of its legacy systems being decommissioned, with the incident still being addressed. According to notification emails sent to affected users, API keys for third-party providers including OpenAI and Anthropic stored in OpenCode Console may have been exposed. The company described the incident as a large-scale distillation attack and said it affected 1.4% of Console users, all of them among users who had connected external providers. It said everyone affected had been notified by email and that users who had not received an email were not affected. OpenCode instructed affected users to revoke and replace their keys and check recent usage for anomalies, and said it was working with Anthropic, OpenAI and Google on remediation. Separately, Anthropic staff said they had revoked Claude API keys they believed were affected and recommended that users who had stored keys in OpenCode create new ones.
Codex fully resets usage limits on day 13 of its 28-day campaign
开发生态
OpenAI’s Codex lead Tibo announced a full reset of usage limits on day 13 of a 28-day initiative, completed at approximately 5 a.m. Beijing time on October 18; the source does not specify the year.
OpenAI’s Codex lead Tibo announced that the team fully reset usage limits on day 13 of its 28-day initiative, completing the reset at approximately 5 a.m. Beijing time on October 18. The source does not specify the year. According to his earlier explanation, the initiative commits the team to one of two actions each day for 28 days: release a clear improvement with practical value for most Codex and Work users, or fully reset usage allowances. The action announced for day 13 falls into the latter category.
Tibo responds to product improvement suggestions: Two are already in progress
开发生态
OpenAI Codex lead Tibo responded on X to five product improvement proposals from user AI Fun, saying that two are in development without identifying them. The reply therefore does not establish whether any particular proposal has entered development.
OpenAI Codex lead Tibo said in a reply to user AI Fun on X that 2 of the user’s 5 product improvement proposals are in development, but did not identify which ones. The proposals cover usage restrictions, feature access, speed, and usage allowances: removing the 5-hour limit for Plus, making dot available to Plus users, changing ultrafast to deliver a 15-fold speed increase, doubling the Pro20x usage allowance, and raising 6.1 Sol’s TPS to at least 50. The reply did not specify which 2 proposals are in development or confirm whether the remaining 3 will be implemented.
WorkBuddy international edition integrates the latest DeepSeek model, free for a limited time
产品应用
WorkBuddy announced that its international version now uses the latest DeepSeek model, with free access through October 31; the source does not specify a year. DeepSeek says DeepSeek-V4.1-Flash-1017 was optimized for platforms including WorkBuddy.
WorkBuddy announced that its international version has been updated to the latest DeepSeek model and will offer free access through October 31; the source gives neither the exact update date nor the year of the free-access deadline. Regarding model compatibility with platforms, DeepSeek said when releasing DeepSeek-V4.1-Flash-1017 that the version was specifically optimized for platforms such as WorkBuddy. The report comes from 橘鸦 AI 早报 and carries a notice stating that the content was created with AI assistance and may contain hallucinations and errors.
