Latest AI News
Stay updated with the latest developments in artificial intelligence
Pricing Aug 21, 2026 1 min readClaude Sonnet 5's "Introductory" Pricing Becomes Permanent
Anthropic priced Claude Sonnet 5 at $2/$10 per 1M tokens at launch, with a scheduled increase to $3/$15 on September 1, 2026. On August 10, it cancelled the increase and made the launch price permanent.
Open Source Aug 18, 2026 1 min readTII's Falcon Perception Does Dense Multimodal Work at 600M Parameters
Abu Dhabi's Technology Innovation Institute extended the Falcon family beyond language with a roughly 600M-parameter model for dense multimodal perception - segmentation, document processing and robotics.
Fact Check Aug 18, 2026 2 min readFact Check: There Is No Llama 5
A detailed "Llama 5, 600B parameters, 5M context, recursive self-improvement" story is circulating. Meta never released it. Here is what actually happened to the Llama line.
Product Launch Aug 18, 2026 1 min readMoonshot Sunsets Older Kimi Models on 31 August
kimi-k2.5 and the moonshot-v1 series are already closed to newly registered users, with a full platform sunset on 31 August 2026 - leaving Kimi K3 as the only option.
Product Launch Aug 18, 2026 1 min readMistral Ships OCR 4 With Self-Hosted Document Extraction
Mistral OCR 4 adds bounding boxes, block classification and inline confidence scores across 170 languages, in a single container for fully self-hosted deployment.
Pricing Aug 16, 2026 2 min readDeepSeek V4-Pro Hits GA and Triples Its Price
DeepSeek's cheapest-frontier-API position just got considerably more expensive: V4-Pro reached general availability on 13 August and the V4 family moved to peak/off-peak billing on 16 August.
Policy Aug 14, 2026 1 min readAnthropic Starts Watermarking Claude Text Output Worldwide
Every Claude model released after 2 August 2026 now embeds an invisible statistical watermark in its word choices, driven by EU AI Act Article 50 - and rolled out globally rather than region by region.
Product Launch Aug 14, 2026 2 min readGLM-5.3 Ships to Coding Plan Subscribers, Weights Held Back
Z.ai's GLM-5.3 reuses the GLM-5.2 base and takes its gains entirely from post-training, with large vendor-reported jumps on agentic coding benchmarks.
Product Launch Aug 13, 2026 1 min readGemini 3.7 Flash Lands at Half the Price of 3.6 Flash
Google shipped Gemini 3.7 Flash roughly three weeks after 3.6 Flash, with a 1M-token context window and introductory pricing that undercuts its predecessor by half - while the Pro line stays delayed.
Open Source Aug 12, 2026 2 min readQwen3.8-Max Is Text-Only — Nearly Everyone Reported Otherwise
Alibaba published the 2.4-trillion-parameter weights on 12 August. The model card contradicts the launch coverage on two counts, and the explanation is a sibling model almost every outlet collapsed into this one.
Product Launch Aug 12, 2026 1 min readGrok 4.6 Keeps Grok 4.5's Foundation and Improves the Training
xAI shipped Grok 4.6 on 7 August 2026 as a 1.5T-parameter successor built on the same V9 foundation as Grok 4.5, with gains from supervised fine-tuning and reinforcement learning rather than a new base model.
Product Launch Aug 12, 2026 1 min readGPT-5.6 Ships in Three Tiers: Sol, Terra and Luna
OpenAI's GPT-5.6 is documented in three tiers sharing a 1.05M token context window, spanning a 25x price range from Sol down to Luna.
Showing 145–156 of 167 stories