OpenAI and Anthropic in price war as Chinese AI rivals gain ground

State of Open Models: Summer 2026 Observations

xAI launches a $120/seat autonomous browser agent the same week DeepSeek gives away a free MIT-licensed agent framework, exposing a closed-vs-open fork in the agentic tooling race.

• xAI's Grok Bot runs on a dedicated cloud instance, logs into browser-based tools like a human (no API needed), and works autonomously 24/7 — $120/seat/month, no free tier, available now on all platforms.
• DeepSeek released Harness v0.1, a free, MIT-licensed, plugin-first agent framework for coding agents — every component (model, tools, memory, UI) is swappable, fully commercial-use allowed.
• OpenAI launched GPT-5.6 Sol Ultrafast via Cerebras hardware: 750 tokens/sec (up to 14x faster) with no intelligence trade-off, enabling real-time voice apps and fast multi-step coding agents.
• OpenAI shipped 'Computer History,' giving ChatGPT persistent memory of your work across different apps.
• A new study shows small, cheap model runs can reliably predict scaling laws if hyperparameters are tuned correctly — useful for budget-constrained model planning.
• Microsoft's Dion3 optimizer cuts training step time 6x versus Muon, a meaningful efficiency gain for anyone training custom models.

Frontier AI is being commoditized overnight: Grok 4.6 and DeepSeek V4-Pro both launched with steep price cuts and long-context agentic capabilities, while Claude's browser sessions now sync across devices.

• xAI launched Grok 4.6 at $2/$6 per million input/output tokens (~60% cheaper than GPT-5.6), with 500K context, function calling, and strong performance on long-running agentic and codebase tasks.
• DeepSeek V4-Pro hit general availability with 1M token context, 384K max output, adjustable reasoning effort, native OpenAI Responses API support, and 50% off-peak pricing for production workloads.
• Anthropic's Claude Chrome extension now syncs as a full Cowork session across desktop, web, and mobile, letting Claude take browser actions (clicking, form-filling) with continuity across devices - but prompt injection risk remains a live concern.
• A new tool can strip invisible watermarks from Claude, Gemini, and OpenAI outputs, raising content-provenance and IP-tracking questions for anyone shipping AI-generated creative work.
• Pathway's 150M parameter model solved ARC-AGI tasks at $0.0007 each, setting a new cost-efficiency benchmark that suggests small specialized models can rival larger ones on narrow reasoning tasks.
• New research shows four specific design choices can degrade long-context performance by up to 47%, a practical warning for anyone building RAG or agent pipelines on these new large-context models.

Google announces Gemini 3.7 Flash just three weeks after previous release

Google undercuts nobody but still ships Gemini 3.7 Flash as enterprises revolt against premium AI pricing, per new Ramp data.

• Google launched Gemini 3.7 Flash at the same price as 3.6 Flash ($0.75/$3.75 per million tokens), still pricier than OpenAI's GPT-5.6 Luna ($0.20/$1.20), but with 10-15pp gains on FrontierCode and DeepSWE coding benchmarks
• Ramp's August index shows Anthropic leads enterprise AI spend (43.5% vs OpenAI's 39.7%), but its priciest model Fable 5 gets only 6% of token usage — businesses are capping spend on frontier-tier models
• OpenAI's GPT-5.6 Sol now has an 'ultrafast' variant running 14x faster, and open-source/Chinese model usage via platforms like Hugging Face grew to 6.1% of Ramp AI spend, alongside xAI's fastest growth since July 2025
• Runway added native integrations with Figma, Dropbox, and Notion to its agent — directly relevant for creative production pipelines
• Perplexity cut cost-per-task by ~10% on its deep research tool via 'Search as Code' optimizations, signaling broader industry pivot toward efficiency over raw capability
• Google's Android chief Sameer Samat outlined a shift from app-tapping to agent-driven task completion across phones, cars, and glasses — a longer-term signal for how creative/consumer interfaces may evolve

Record, train, and deploy from one place with Strands Agents, LeRobot, and Hugging Face Storage Buckets

OpenAI embeds itself inside IBM Consulting while Alibaba and xAI quietly ship stronger creative-generation models.

• OpenAI-IBM deal creates an 'OpenAI Practice' inside IBM Consulting with thousands of certified consultants to deploy GPT-5.6, Codex, and ChatGPT Work into legacy enterprise systems (finance, HR, security via Project Daybreak).
• Google's Pixel 11 pushes an 'agentic phone' vision — sandboxed multi-step task agents, Rambler voice keyboard, and Magic Capture — while deliberately avoiding the 'AI' label in marketing.
• Mozilla's new CEO is positioning Firefox as model-agnostic (user picks their own AI model in Smart Window), explicitly betting on open-weight ecosystems over building a frontier model.
• Alibaba shipped Wan-Animate-2, a high-fidelity real-time character animation model — directly relevant for creative/video pipelines.
• xAI released Grok 4.6 with improved long-running agent performance and more capable interactive/visual generation.
• Anthropic launched Claude Cowork in Chrome's side panel for cross-tab browser agent work; OpenAI added sync for ChatGPT Work/Codex projects, chats, and plugins.

Claude's new Scarlet Letter watermark is invisible—for now

Don't classify. Hallucinate!

• <p><strong><a href="https://softwaredoug.com/blog/2026/08/10/hypothetical-classifications">Don&#x27;t classify. Hallucinate!</a></strong></p> I still have quite a bit of older content on my blog that I never got round to tagging. My blog has <a href="https://simonwillison.net/">1,856 tags</a> - like

sqlite-utils 4.2

• <p><strong>Release:</strong> <a href="https://github.com/simonw/sqlite-utils/releases/tag/4.2">sqlite-utils 4.2</a></p> <p>Lots of improvements in this one relating to the <a href="https://sqlite-utils.datasette.io/en/stable/python-api.html#transforming-a-table">table.transform() feature</a>, which

llm-gemini 0.33

• <p><strong>Release:</strong> <a href="https://github.com/simonw/llm-gemini/releases/tag/0.33">llm-gemini 0.33</a></p> <p>It's been a while since the last <code>llm-gemini</code> release. This version of the plugin adds support for today's <a href="https://blog.google/innovation-and-ai/models-and-res