https://botsdothat.com/news/ New AI models, agent launches and price changes, explained in plain words.en Wed, 21 Oct 2026 12:00:00 +0000https://botsdothat.com/news/#anthropic-claude-duetanthropic-claude-duet Wed, 21 Oct 2026 12:00:00 +0000agent launchClaude Duet sends you a morning brief from your connected apps, keeps an eye on them through the day, works on tasks while you're away, and asks before it does anything big like paying or deleting. It's a waitlist research preview for US Pro and Max subscribers, on iOS, Android and claude.ai, and Anthropic says it can still misjudge what's urgent. Source: Anthropic https://claude.com/blog/claude-duethttps://botsdothat.com/news/#openai-patch-the-planet-phpopenai-patch-the-planet-php Wed, 21 Oct 2026 12:00:00 +0000industryIf you run a website or blog on PHP (WordPress included), make sure your host has updated PHP to the patched version now. OpenAI says its models helped researchers find the bugs before attackers did, and it's sharing a version checker and a Codex skill to help site owners check. Source: OpenAI https://openai.com/index/coordinated-disclosure-of-critical-vulnerabilitieshttps://botsdothat.com/news/#claude-security-continuous-scansclaude-security-continuous-scans Wed, 21 Oct 2026 12:00:00 +0000featureTeams on Claude Enterprise can connect their code once and Claude keeps checking new changes for security holes, marking each one fixed when the fix lands. It's a public beta for businesses, not something personal plans get. Source: Anthropic https://claude.com/resources/articles/claude-security-now-scans-continuously-and-tracks-fixeshttps://botsdothat.com/news/#openai-models-on-google-cloudopenai-models-on-google-cloud Wed, 21 Oct 2026 12:00:00 +0000industryCompanies on Google Cloud will be able to use GPT-6 models and Codex without a separate OpenAI account (private preview now, wider launch planned in the coming weeks). For you it mostly means more places your work tools can quietly plug in OpenAI models. Source: OpenAI https://openai.com/index/openai-models-google-cloudhttps://botsdothat.com/news/#mistral-voxtral-3mistral-voxtral-3 Tue, 20 Oct 2026 12:00:00 +0000model releaseA new set of speech models plus a kit for building phone and web voice agents that listen, look things up and talk back in 50+ languages. Mistral lists its voice agent API at $0.04 a minute with Mistral Small 4, so phone bots keep getting cheaper to run. Source: Mistral https://mistral.ai/news/voxtral-3/https://botsdothat.com/news/#openai-12-hour-prompt-cachingopenai-12-hour-prompt-caching Tue, 20 Oct 2026 12:00:00 +0000pricingBots that run hours apart (like a morning and an evening check) can now keep their fixed instructions cached and pay the lower cached-input price. Writing to the 12-hour cache costs 2.5x the normal input price, so it pays off when the same setup is reused several times. Source: OpenAI API changelog https://developers.openai.com/api/docs/changeloghttps://botsdothat.com/news/#deepseek-deepinferdeepseek-deepinfer Tue, 20 Oct 2026 12:00:00 +0000industryDeepInfer is the software DeepSeek uses to serve V4.1 Flash cheaply, now free under Apache 2.0. Hosting companies can copy the tricks, which could push the price of running open models down for your bots. Source: DeepSeek (GitHub) https://github.com/deepseek-ai/DeepInferhttps://botsdothat.com/news/#android-police-gems-become-skillsandroid-police-gems-become-skills Sun, 18 Oct 2026 12:00:00 +0000featureFrom November 17, Google starts converting personal Gems into Skills you call with a slash in the main chat, and Scheduled Actions no longer need a paid plan (still rolling out to free accounts). If you built a Gem-based bot, test it before the switch. Source: Android Police https://www.androidpolice.com/googles-renamed-gemini-features-field-guide/https://botsdothat.com/news/#deepseek-v4-1-flash-1017deepseek-v4-1-flash-1017 Sat, 17 Oct 2026 12:00:00 +0000model releaseDeepSeek says the update finishes tasks faster with shorter outputs, which means fewer tokens to pay for. The API name and prices are unchanged, and the open weights are on Hugging Face. Source: DeepSeek API changelog https://api-docs.deepseek.com/updates/https://botsdothat.com/news/#deepseek-search-fetch-apideepseek-search-fetch-api Sat, 17 Oct 2026 12:00:00 +0000pricingBots built on DeepSeek can now search the web and read pages through DeepSeek's own API at $0.75 per 1,000 requests each. Calls made through some coding agents are free until billing starts on November 1 (Beijing time). Source: DeepSeek pricing https://api-docs.deepseek.com/quick_start/pricinghttps://botsdothat.com/news/#meta-muse-spark-openmeta-muse-spark-open Thu, 15 Oct 2026 12:00:00 +0000model releaseMeta's new open model is built for coding and agent work, and you can call it on day one through Groq, OpenRouter, Together AI and other APIs. Groq runs it at 350+ tokens per second per user. That speed matters for agents, because each task is a long chain of model calls. Source: Meta https://research.meta.ai/blog/muse-spark-openhttps://botsdothat.com/news/#grok-voice-3grok-voice-3 Wed, 14 Oct 2026 12:00:00 +0000model releaseThe voice model behind Grok Bot's voice mode and Tesla's in-car assistant got an upgrade, at the same $0.08 a minute on the API. xAI says it answers in under a second and handles interruptions, which is what makes a phone bot feel natural. Source: xAI https://x.ai/news/grok-voice-3https://botsdothat.com/news/#gemini-4-argon-availablegemini-4-argon-available Tue, 13 Oct 2026 12:00:00 +0000model releaseGoogle's new top model is in the API at an introductory $2 / $10 per million tokens, and in the Gemini app for Google AI Pro and Ultra subscribers. Google built it for long, multi-step work, the kind agents do. Source: Google https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/https://botsdothat.com/news/#meta-muse-wondermeta-muse-wonder Mon, 12 Oct 2026 12:00:00 +0000model releaseMeta says Muse Wonder will soon power its Muse personal agent, starting with paying subscribers. Developers can use it through Meta's API at $4 / $20 per million tokens. Source: Meta https://dev.meta.ai/resources/blog/build-with-muse-wonderhttps://botsdothat.com/news/#gemini-spark-becomes-taskgemini-spark-becomes-task Mon, 12 Oct 2026 12:00:00 +0000featureHanding a job to Google's background agent is now one tap away in the Gemini app, with a new Scheduled page that shows active, paused and finished tasks. It's live on Android and iOS; web wasn't confirmed yet. Source: 9to5Google https://9to5google.com/2026/10/12/gemini-spark-task-redesign/https://botsdothat.com/news/#openai-agents-sandbox-billingopenai-agents-sandbox-billing Fri, 09 Oct 2026 12:00:00 +0000pricingAgents built on OpenAI's Agents API run in a hosted sandbox computer, and that is now billed only during active turns, not while the agent waits for you. Sandboxes can also be pre-warmed so replies start faster. Source: OpenAI API changelog https://developers.openai.com/api/docs/changeloghttps://botsdothat.com/news/#claude-dashboards-motionclaude-dashboards-motion Thu, 08 Oct 2026 12:00:00 +0000featureAsk a question in plain words and Claude builds a dashboard that stays current, showing the query behind every number. Motion turns a report into a short animation you can export as MP4. Both are in beta on paid plans (Motion on Team and Enterprise). Source: Anthropic https://claude.com/resources/articles/dashboards-and-motionhttps://botsdothat.com/news/#gpt-6-for-everyonegpt-6-for-everyone Wed, 07 Oct 2026 12:00:00 +0000model releaseChatGPT answers can now include charts, buttons, forms and small tools you use right in the chat. Paid plans get GPT-6 Sol and Free and Go users get GPT-6 Luna, starting October 8. Source: OpenAI https://openai.com/index/gpt-6-for-everyone/https://botsdothat.com/news/#claude-haiku-5-5claude-haiku-5-5 Wed, 07 Oct 2026 12:00:00 +0000model releaseAt $0.10 / $0.50 per million tokens for prompts up to 100,000 tokens, Anthropic says it costs about 75% less to run than Haiku 4.5. It's built for summaries, sorting and browser use: the simple, repeated jobs most everyday bots do. Source: Anthropic https://www.anthropic.com/news/claude-haiku-5-5https://botsdothat.com/news/#claude-sonnet-cache-cut-and-creditsclaude-sonnet-cache-cut-and-credits Wed, 07 Oct 2026 12:00:00 +0000pricingCached reads on Claude Sonnet 5.5 dropped from $0.20 to $0.10 per million tokens, which Anthropic says makes most agent work about 20% cheaper. Max subscribers get $100 or $200 a month in API credits, and Team plans up to $500, to build their own bots. Source: Anthropic https://www.anthropic.com/news/claude-haiku-5-5https://botsdothat.com/news/#mistral-large-4-previewmistral-large-4-preview Tue, 06 Oct 2026 12:00:00 +0000model releaseMistral's biggest model yet is in a public preview on its API, and Mistral says anyone will be able to download and run it themselves later this month. That's good news if you want a strong model you can host privately. Source: Mistral https://mistral.ai/news/mistral-large-4https://botsdothat.com/news/#claude-google-workspaceclaude-google-workspace Tue, 06 Oct 2026 12:00:00 +0000featureA Claude sidebar can read the file you have open and edit it in place, asking before each change by default. It's in beta on all paid Claude plans, so your doc-and-spreadsheet chores can go to Claude without copy-pasting. Source: Anthropic https://claude.com/resources/articles/claude-now-works-in-google-docs-sheets-and-slideshttps://botsdothat.com/news/#gpt-6-1-solgpt-6-1-sol Tue, 29 Sep 2026 12:00:00 +0000model releaseOpenAI says GPT-6.1 Sol nearly matches its top model, GPT-6 Astra, on agent work at one-fifth of Astra's token prices ($2 / $10 per million). For bot builders, it's a cheaper middle option for jobs a small model can't handle. Source: OpenAI https://openai.com/index/introducing-gpt-6-1-sol/https://botsdothat.com/news/#openai-agents-api-computer-useopenai-agents-api-computer-use Tue, 29 Sep 2026 12:00:00 +0000featureAgents built on OpenAI's platform can now complete tasks in an OpenAI-hosted browser, with the app asking you before it visits a site and handling sign-in. Expect more third-party bots that click through websites for you. Source: OpenAI API changelog https://developers.openai.com/api/docs/changeloghttps://botsdothat.com/news/#openai-dotsopenai-dots Tue, 29 Sep 2026 12:00:00 +0000agent launchA dot runs on GPT-6 Astra with its own cloud computer, connects to your apps, and keeps working between chats, checking with you on decisions. It's rolling out to ChatGPT Pro and Business Premium users first (not the EU, Switzerland or the UK on Pro at launch). Source: OpenAI https://openai.com/index/introducing-dots/https://botsdothat.com/news/#claude-sonnet-5-5claude-sonnet-5-5 Mon, 28 Sep 2026 12:00:00 +0000model releaseAnthropic says Sonnet 5.5 runs 30%+ faster than Sonnet 5 and costs up to 30% less for most work. It's the middle model many agent apps use by default. Source: Anthropic https://www.anthropic.com/news/claude-sonnet-5-5https://botsdothat.com/news/#grok-bot-team-botsgrok-bot-team-bots Mon, 28 Sep 2026 12:00:00 +0000agent launchOne person sets up a bot with the files, apps and skills it needs, then the whole team can use it in Slack or Grok Bot, while each person's chats stay private. It's in public beta for Teams and Enterprise plans. Source: xAI https://x.ai/news/team-botshttps://botsdothat.com/news/#claude-opus-5-5claude-opus-5-5 Tue, 22 Sep 2026 12:00:00 +0000model releaseAnthropic says Opus 5.5 leads in agentic coding and knowledge work and costs 40% less to run than Opus 5 on typical workloads. It's the model to reach for when a bot's job is hard, not routine. Source: Anthropic https://www.anthropic.com/news/claude-opus-5-5https://botsdothat.com/news/#gpt-6-sol-and-lunagpt-6-sol-and-luna Tue, 22 Sep 2026 12:00:00 +0000model releaseTwo new models: Sol for most work and Luna, a very cheap one at $0.10 / $0.50 per million tokens. Luna is the kind of model to use for simple, frequent bot jobs. Source: OpenAI API changelog https://developers.openai.com/api/docs/changeloghttps://botsdothat.com/news/#grok-4-7grok-4-7 Mon, 21 Sep 2026 12:00:00 +0000model releasexAI's new model for coding and knowledge work keeps the same $2 / $6 per million token price as Grok 4.6. xAI says it was trained to work inside the Grok Bot harness, which should make Grok Bot better at everyday tasks. Source: xAI https://x.ai/news/grok-4-7