DeepSeek released build V4-Pro-0813 with major agent benchmark gains, open-sourced its Deepseek Harness agent framework under MIT license, and announced API price increases of up to 52% effective August 16.
DeepSeek has released DeepSeek-V4-Pro-0813, a 1.7-trillion-parameter model that supersedes the DeepSeek-V4-Pro preview. The model adds a DSpark speculative decoding module and posts measurable gains on agentic and coding benchmarks, according to DeepSeek's technical report.
Amazon has released Microsoft 365 extensions for its Quick AI assistant, embedding agentic capabilities directly into Word, Excel, PowerPoint, and Outlook. The extensions run entirely in the cloud, require no client-side installation, and connect to existing Quick data sources like Salesforce, Jira, Slack, and SharePoint.
Microsoft is merging its consumer-facing Copilot app with Microsoft 365 Copilot while discontinuing Group Chats, AI-generated podcasts, Copilot Labs, Deep Research, and the Mico animated character by August 18, 2026. Paying professional users will get Researcher as a Deep Research replacement.
Ant Group's inclusionAI released Ling 3.0 Flash, which scores 38 points on the Artificial Analysis Intelligence Index — the highest of any open model under 124 billion total parameters. The model cuts its hallucination rate from 97 to 44 percent versus its predecessor and ships under an MIT license.
Microsoft will merge its consumer Copilot app and Microsoft 365 Copilot into a single unified app starting mid-September 2026, letting users sign in with personal, work, or school accounts. Three features—group chats, podcasts, and Deep Research—will be retired on August 18.
New Ramp spending data shows Anthropic's flagship Fable 5 model, priced at $10/$50 per million tokens, is seeing weak corporate adoption compared to OpenAI's GPT-5.6 Sol. Analysts suggest frontier AI pricing may have hit a ceiling.
IAPS fellow Severin Field interviewed 25 researchers from top AI labs about recursive self-improvement in late 2025. Several milestones they cited as evidence of progress — Math Olympiad gold, autonomous training loops, majority AI-written code — have since occurred, according to a new report.
Anthropic's Chrome extension now runs full Claude Cowork sessions in its side panel, letting skills, plugins, and connectors operate directly in the browser. The update is live for all paid plans via the Chrome Web Store.
xAI's Grok 4.6 scored nearly even with OpenAI's GPT-5.6 Sol Max on the Artificial Analysis Intelligence Index, while Meta launched Muse Glimmer, a laptop-runnable open-weight model. Musk says Grok 4.7 will arrive in three to four weeks and claims it will 'exceed all current models.'
Cline's CLI v3.0.54 release fixes a major bug that made the Claude Code provider unusable for agentic work, along with tool-call JSON truncation handling and token telemetry that was overcounting cache-heavy sessions by roughly 5x. The update also changes how Managed Hub daemons handle version conflicts between multiple Cline installs.
xAI released Grok 4.6, a confirmed 1.5T-parameter model built on Grok 4.5 with heavier training on long-horizon agentic tasks. It powers the newly launched Grok Bot product and scores 61 on Artificial Analysis's Intelligence Index at $2/$6 per 1M input/output tokens — well below frontier competitors.
Hugging Face repositories for Qwen3.8-2.4T-A95B, its FP8 variant, and NVIDIA's Nemotron-3.5-Lightning-30B-A3B have surfaced, but neither company has published accompanying benchmarks, technical reports, or pricing.
Alibaba's Qwen team has released Qwen3.8-2.4T-A95B-FP8, an open-weight, FP8-quantized MoE model with 2.4 trillion total parameters and 95 billion activated per token. It natively supports 262,144 tokens of context, extensible to 1,010,000, and forms the base for the hosted Qwen3.8-Max API.
Anthropic has integrated Claude Cowork into its Claude for Chrome browser extension, letting users continue tasks between the desktop app and browser with shared history and connectors. Max and Team subscribers get access immediately, with Pro plan support coming in the following weeks.
Anthropic has begun embedding invisible watermarks in Claude's text outputs to comply with the EU AI Act's Transparency Code. The move has triggered backlash from some users worried the watermarks will expose their undisclosed use of AI at work or in school.
NVIDIA released the full-precision BF16 reference weights for Nemotron 3.5 Lightning, a 30B-parameter Mixture-of-Experts model with only 3B active parameters and support for up to 1 million tokens of context. The model uses a hybrid Mamba-2, MoE, and Attention architecture and is licensed under OpenMDW-1.1 for commercial use.
Anthropic has upgraded Claude's Chrome side panel into a full Cowork session, letting browser-based conversations and tasks carry over to Claude's desktop, web, and mobile apps. The feature is live now for Max and Team customers, with Pro access rolling out in coming weeks.
xAI's Grok 4.6 ties OpenAI's GPT-5.6 Sol on the Artificial Analysis Intelligence Index with a score of 61, trailing only Anthropic's Claude Opus 5 and Claude Fable 5. Pricing remains at $2/$6 per million tokens, undercutting both competitors by more than 60 percent.
Raycast v0.71 introduces Screen Awareness, a feature that feeds context from the active app or window—including selected text, app details, and a screenshot—directly into AI Chat or Quick AI. The update also adds context management, dictation vocabulary import, and new keyboard chord support.