OpenAI has introduced GPT-6 Sol and GPT-6 Luna, expanding the GPT-6 universe to distribute the benefits of GPT-6 Astra's intelligence more broadly. These new models were trained using similar methods to GPT-6 Astra, bringing state-of-the-art performance in professional work, factuality, coding, computer use, and alignment to faster, more affordable options.

A significant improvement is the 50% reduction in API prices for GPT-6 Sol and Luna compared to their GPT-5.6 promotional pricing, making advanced AI more practical for everyday tasks. GPT-6 Sol, for instance, significantly outperforms Claude Opus 5 on AutomationBench and achieves high scores on Agents' Last Exam at a fraction of the cost.

The models also boast enhanced coding capabilities, demonstrating strong performance on benchmarks like FrontierCode and DeepSWE, often matching or exceeding competitors at much lower costs. Furthermore, their collaboration style has been refined, leading to clearer, more concise, and less jargon-filled responses.

Improvements in caching for agents and long conversations are also a key feature, with enhanced prompt caching delivering higher cache hit rates and offering 90% discounts on cached input-token reads. This allows developers to build more efficient applications.

In terms of alignment, both Sol and Luna show improvements over their GPT-5.6 counterparts, including lower rates of misleading claims in coding work. These models are available in ChatGPT Work and Codex, and via the OpenAI API as gpt-6-sol and gpt-6-luna.

👉 Read the full article on OpenAI

Anthropic has introduced Claude Opus 5.5, the first model in its new Claude 5.5 family. This model performs at the level of Claude Fable 5.1 on most tasks while costing 40% less to run than Opus 5.

Opus 5.5 is positioned as a leading model in agentic coding, computer use, and knowledge work, demonstrating significant performance jumps in complex tasks such as large-scale code migrations and audits. One tester, for example, completed a 680,000-line code migration in less than a day.

Safety has been substantially improved, with Opus 5.5 achieving the best scores to date on Anthropic's automated behavioral audit and showing increased resistance to prompt injection. Strict safeguards, similar to those on Claude Fable 5.1, are applied in high-risk areas like cybersecurity and biology.

In terms of cost and speed, Opus 5.5 requires less compute than Opus 5, with input tokens priced at $4 per million and output tokens at $20 per million, representing a 20% reduction from Opus 5.

Communication has also seen major improvements, with Opus 5.5 generating clearer and easier-to-follow writing, which enhances collaboration during long working sessions. Claude Sonnet 5.5 and Claude Haiku 5.5 are expected to follow in the coming weeks with similar enhancements.

👉 Read the full article on Anthropic

Google has partnered with the Gates Foundation to commit $100 million towards providing AI-powered climate and crop insights to 200 million smallholder farmers across the Global South. This initiative aims to bridge the gap for farmers who often lack access to satellite data, financial services, and public support, especially in the face of increasingly unpredictable weather.

The combined funding and dedicated technical support from Google researchers will facilitate the deployment of real-time climate, agricultural, and language AI tools. This will enable farmers in Sub-Saharan Africa and South Asia to manage their crops more effectively and adapt to climate change.

This partnership significantly expands the reach from a previous target of 50 million farmers to 200 million, marking a crucial step in leveraging AI to improve the livelihoods and productivity of smallholder farmers who produce nearly 35% of the world's food.

👉 Read the full article on Google Blog