OpenAI Launches GPT-6.1 Sol Ultrafast Version with Up to 8x Speed Improvement

OpenAI Developers officially announced the launch of the Ultrafast version of GPT-6.1Sol in the API, Codex, and ChatGPT Work platforms.

As an upgraded performance version, this version can run up to 8 times faster than the standard Sol version. In terms of pricing, it is 6 times the cost of the standard version. The specific charges are $12 (approximately 80.5 CNY) per million input tokens and $60 (approximately 402.7 CNY) per million output tokens.

In terms of applicable scope and permission access, this new feature has been officially launched in Codex and ChatGPT Work. It is available for Pro500 users, qualified pay-as-you-go enterprise plans, and education plans with quota-based billing. Enterprise users need to have their administrators activate the relevant permissions in the background.

Currently, the Ultrafast version of GPT-6.1Sol has been deployed in all supported regions and fully supports data residency compliance services in the United States and the European Union.

OpenAI Collaborates with Cerebras to Launch Ultrafast Mode, GPT-5.6Sol Speeds Up by 14 Times

On August 13, OpenAI released the "Ultrafast" mode, which is integrated into the flagship large model GPT-5.6Sol. The speed is 14 times faster than standard processing, generating up to 750 output tokens per second while maintaining the original intelligence and output quality. This feature was jointly developed by OpenAI and chip manufacturer Cerebras, and is currently in preview stage, initially available to select enterprise customers, with further expansion expected as computing power increases.

An infinite canvas to hold all creations: Douba Work adds new features for images, videos, and HTML to be integrated into the same canvas

On October 9th, Douba announced a dual-line update for 'Douba Work': a new canvas function has been added to the work mode, integrating the new image model Seedream5.0Flash; Douba 2.1Lite (version 0921) is also launched, focusing on saving quotas and fast delivery. The canvas is designed for complex creation, allowing users to view, compare, and organize materials, proposals, and results on the same infinite canvas. After generation, text, color schemes, and layout can be adjusted, editing from the overall to the details.

JetBrains Releases Programming AI Model Mellum2.1 with Twice the Throughput of Competitors Under Heavy Load

JetBrains launches a new programming AI model called Mellum2.1, focusing on enhancing agent-based programming capabilities. The model continues with a 12B mixture-of-experts architecture and open-source licensing, lowering the barrier for enterprise and individual developers to deploy privately. In terms of training, reinforcement learning has expanded from short-term completion to become the core training component, with specialized reinforcement in millions of sandbox environments for software engineering and algorithms.

Covering All Family AI Needs? Microsoft May Allow Microsoft 365 Family Members to Share Copilot Quotas

Microsoft is reportedly planning to adjust the distribution of Copilot benefits, with plans to remove the restriction that only the primary subscriber can use it. This would allow Microsoft 365 Family and Premium primary subscribers to share Copilot and AI features with up to five family members. According to Neowin, after the adjustment, the primary subscriber would also be able to manage AI access for minor family members. Currently, Microsoft's official support page still states that this feature is reserved for the primary subscriber.

OpenAI Launches GPT-6.1 Sol Ultrafast Version with Up to 8x Speed Improvement

Related Recommendations

OpenAI Collaborates with Cerebras to Launch Ultrafast Mode, GPT-5.6Sol Speeds Up by 14 Times

An infinite canvas to hold all creations: Douba Work adds new features for images, videos, and HTML to be integrated into the same canvas

Anthropic Launches OSS Scanner, Free Vulnerability Scanning for Global Open Source Software with the Strongest AI Model

JetBrains Releases Programming AI Model Mellum2.1 with Twice the Throughput of Competitors Under Heavy Load

Covering All Family AI Needs? Microsoft May Allow Microsoft 365 Family Members to Share Copilot Quotas