https://www.insighton.ai/anthropic/rss.xmlIndependent Anthropic news and analysis from InsightOn.ai.https://www.insighton.ai/anthropic/stories/haiku-55/https://www.insighton.ai/anthropic/stories/haiku-55/Wed, 07 Oct 2026 11:00:00 GMTAnthropic released Claude Haiku 5.5 on 7 October, cutting the entry price for applications that need inexpensive, frequent model responses. Its direct API charges US$0.10 per million input tokens and US$0.50 per million output tokens for prompts of up to 100,000 tokens, one tenth of Haiku 4.5’s corresponding rates. Longer prompts attract higher prices, making the size of the material supplied to the model an important purchasing variable. The release extends Anthropic’s latest model generation into workloads where repeated small charges can determine whether an application is economical. Buyers gain a cheaper option for routine requests, while developers still have to establish whether it completes their particular tasks reliably enough to reduce total costs.https://www.insighton.ai/anthropic/stories/sonnet-cache/https://www.insighton.ai/anthropic/stories/sonnet-cache/Wed, 07 Oct 2026 11:00:00 GMTAnthropic cut Claude Sonnet 5.5’s cached-input price to US$0.10 per million tokens on 7 October, reducing the cost of repeatedly supplying material the model has already processed. The previous rate was US$0.20, while ordinary input and output prices remain US$2 and US$10 per million tokens respectively. The change targets a different part of application economics from a general model-price reduction: it benefits requests that successfully reuse cached context. For customers running agents against recurring instructions, reference documents or other stable material, the saving depends on how much of the bill comes from cache reads. Applications dominated by fresh input or generated output will receive a much smaller percentage reduction in their total spend.https://www.insighton.ai/anthropic/stories/api-credits/https://www.insighton.ai/anthropic/stories/api-credits/Wed, 07 Oct 2026 11:00:00 GMTAnthropic announced monthly API credits for Claude Max and Team subscriptions on 7 October, with rollout planned during the week, giving existing subscribers a funded route into its developer platform. Max 5x plans receive US$100 a month and Max 20x plans US$200, while Team credits are pooled and capped at US$500 per organisation. The credits are separate from the interactive subscription allowance and apply to eligible direct-platform usage. The commercial consequence is a lower immediate cash hurdle for subscribers trying an application, automated workflow or agent through the API. This is a recurring usage allowance, not cash paid to customers, and its value depends on whether they use the eligible services before the balance expires.https://www.insighton.ai/anthropic/stories/cyber-verification/https://www.insighton.ai/anthropic/stories/cyber-verification/Tue, 06 Oct 2026 11:00:00 GMTAnthropic expanded its Cyber Verification Program on 6 October, organising access to advanced model capabilities into three levels for defensive work, authorised red teaming and specialised critical-system testing. The programme covers Opus 5.5, Sonnet 5.5 and Mythos 5.1, with eligibility and review requirements varying by the work involved. It turns access to sensitive capabilities into a more explicit operating arrangement for security customers rather than a uniform permission granted with a model subscription. The commercial opportunity lies in letting legitimate teams use more capable tools on demanding work, while the conditions on access create a separate deployment decision involving the organisation’s identity, intended tasks and willingness to accept the programme’s monitoring requirements.https://www.insighton.ai/anthropic/stories/google-workspace/https://www.insighton.ai/anthropic/stories/google-workspace/Tue, 06 Oct 2026 11:00:00 GMTAnthropic released a public beta on 6 October that lets paying Claude users edit Google Docs, Sheets and Slides directly, bringing its assistant into files where office work already happens. A Workspace add-on places Claude beside an open document, while separate connectors let users create and edit Google files from a Claude conversation. The two routes reduce the need to copy material between a chat window and a productivity application. For Anthropic, this is a distribution move as well as a feature release: customers can use its models inside Google’s working environment. Availability does not establish productivity gains, but it puts those gains within the document workflow rather than requiring a separate destination.https://www.insighton.ai/anthropic/stories/booz-allen/https://www.insighton.ai/anthropic/stories/booz-allen/Tue, 06 Oct 2026 11:00:00 GMTBooz Allen Hamilton reported that one analyst using Anthropic’s Claude Mythos assessed eight production systems spanning 138 repositories in 12 days, describing work that it estimated would otherwise take a larger team months. The result appears in an Anthropic customer account published on 6 October and supplies a concrete scope and duration for a security application. It is a customer estimate of the alternative effort, rather than a controlled comparison between matched teams. Even with that limitation, the case identifies a potential operating change: a specialist can investigate relationships across a substantial code estate while using the model to extend the breadth of the review, leaving validation and consequential decisions with human practitioners.https://www.insighton.ai/anthropic/stories/broadcom-financing/https://www.insighton.ai/anthropic/stories/broadcom-financing/Fri, 02 Oct 2026 11:00:00 GMTBroadcom has begun assembling a reported US$60 billion financing package for chips serving Anthropic and other artificial-intelligence customers, according to Bloomberg’s 2 October account. The proposed structure comprises US$42 billion of senior secured financing and US$18 billion of junior financing, with Blackstone expected to lead the latter component. The report describes preparations and prospective commitments, not a completed fundraise or money already received by Anthropic. Its significance is the effort to bring outside capital into the infrastructure supporting model demand. Chip supply alone does not fund an expanding compute estate; lenders and investors must also be willing to finance the assets and accept the risks attached to future customer payments.https://www.insighton.ai/anthropic/stories/frontier-academy/https://www.insighton.ai/anthropic/stories/frontier-academy/Fri, 02 Oct 2026 11:00:00 GMTAnthropic committed US$100 million to Claude Frontier Academy on 2 October, aiming to train 10,000 engineers by the end of 2027 in building practical enterprise systems with Claude. The first cohorts are already running in San Francisco, New York and London, with participants from organisations including Accenture, Deloitte, Morgan Stanley and Novo Nordisk. The programme combines intensive instruction with work on a real organisational project. It addresses a commercial constraint that a more capable model does not automatically solve: customers need people able to turn an intended use into a working system. The commitment is a programme investment, not a stated cash payment to each participant or evidence that every proposed project will reach production.https://www.insighton.ai/anthropic/stories/broadcom-facility/https://www.insighton.ai/anthropic/stories/broadcom-facility/Thu, 01 Oct 2026 11:00:00 GMTAnthropic’s confidential listing documents identify it as the customer eligible for up to US$42 billion of Broadcom convertible-note financing linked to a US$125.2 billion, five-year chip lease, Reuters reported on 1 October. Broadcom had previously disclosed an unnamed customer facility in its public filing. The identification connects a major financing provision directly to Anthropic’s infrastructure obligations, but it does not mean the company has drawn the full amount. Reuters reported that no notes were expected to be sold before the planned flotation. The arrangement could help finance access to compute while tying the source of funding closely to the supplier relationship that creates the underlying payment obligation.https://www.insighton.ai/anthropic/stories/barclays/https://www.insighton.ai/anthropic/stories/barclays/Thu, 01 Oct 2026 11:00:00 GMTBarclays expanded its collaboration with Anthropic on 1 October, describing operating uses that include processing 120,000 Global Markets emails a day and a knowledge assistant used by more than 16,000 colleagues. The bank also plans to bring Claude Code to half of its developers by the end of 2026 and a majority during 2027. The combination separates current workflows from a future software rollout: the email and knowledge applications are described as operating uses, while wider coding access remains a target. For Anthropic, the agreement offers several routes into the same enterprise, with model consumption and adoption potentially spreading across information retrieval, message handling and the development of the bank’s own systems.https://www.insighton.ai/anthropic/stories/cloud-rivals/https://www.insighton.ai/anthropic/stories/cloud-rivals/Tue, 29 Sep 2026 11:00:00 GMTAnthropic routed 47% of its 2025 sales through Amazon and Google cloud marketplaces, according to figures in a confidential IPO prospectus reviewed by Reuters and reported on 29 September. The $2.16 billion channel made two companies simultaneously major distributors, compute suppliers, investors and AI competitors. Reuters calculated that Anthropic paid about $351 million in marketplace distribution fees, roughly 16 cents for each dollar sold there. Two unnamed customers each supplied another 12% of total revenue. The figures reveal how quickly Claude has reached enterprise buyers through incumbent clouds, while showing that a meaningful share of Anthropic's customer access and cash collection runs through partners with their own strategic interests.https://www.insighton.ai/anthropic/stories/founder-control/https://www.insighton.ai/anthropic/stories/founder-control/Tue, 29 Sep 2026 11:00:00 GMTAnthropic's seven co-founders would direct 50.1% of shareholder votes through a proposed Founder LLC and a single powerful Class F share after its IPO, according to prospectus details reported by Reuters on 29 September. The arrangement covers major corporate matters including some board elections. Ordinary Class A shareholders would have one vote per share but little practical ability to outvote the founder bloc. Anthropic says the structure is designed to protect its public-benefit mission as it enters markets that may press for faster model releases and greater returns on enormous compute commitments. It is an unusually concrete answer to who will decide when commercial opportunity conflicts with the lab's stated safety priorities.https://www.insighton.ai/anthropic/stories/white-house-accord/https://www.insighton.ai/anthropic/stories/white-house-accord/Tue, 29 Sep 2026 11:00:00 GMTAnthropic chief executive Dario Amodei signed a voluntary frontier AI safety accord at the White House on 29 September alongside leaders of other major AI companies. The agreement calls for robust internal controls, an independent external auditor and a board committee to review findings, while leaving open the possibility that its measures could later become law. President Donald Trump described the arrangement as industry self-policing. Amodei, who has publicly urged a slower pace for the most capable systems, said at the event that the technology has ‘very real risks.’ His signature brings Anthropic's safety stance into a shared industry commitment at a moment when its models and IPO plans are attracting scrutiny.https://www.insighton.ai/anthropic/stories/compute-plan/https://www.insighton.ai/anthropic/stories/compute-plan/Mon, 28 Sep 2026 11:00:00 GMTAnthropic's confidential IPO prospectus, reviewed by Reuters and first described on 28 September, sets out at least $518 billion of cloud, compute and infrastructure spending over the coming decade. About four-fifths of the amount is non-cancelable or payable regardless of use, according to a further Reuters examination of the document. The Claude developer generated nearly $4.6 billion in 2025 revenue, about 12 times the prior year, but lost more than $8 billion from operations and spent $7.33 billion on compute and infrastructure. Its nearly $42 billion net loss includes a roughly $34 billion non-cash charge linked to financing value. The figures make the proposed public offering a judgment on whether extraordinary future demand can earn a return on capacity committed years in advance.https://www.insighton.ai/anthropic/stories/sonnet-55/https://www.insighton.ai/anthropic/stories/sonnet-55/Mon, 28 Sep 2026 11:00:00 GMTAnthropic released Claude Sonnet 5.5 on 28 September, positioning it as a faster, more economical default for coding and everyday professional work. It generates output more than 30% faster than Sonnet 5 and costs up to 30% less per task in Anthropic's tests while retaining the same $2 per million input tokens and $10 per million output tokens. The improvement comes from doing work with fewer tokens and steps, not a cut in the headline API rate. Sonnet 5.5 is the second model in the 5.5 family after Opus 5.5. Anthropic says it approaches the stronger model on some structured tasks, allowing buyers to reserve the more expensive Opus for complex judgment and open-ended work.https://www.insighton.ai/anthropic/stories/pentagon-blacklist/https://www.insighton.ai/anthropic/stories/pentagon-blacklist/Fri, 25 Sep 2026 11:00:00 GMTA divided federal appeals court on 25 September allowed the Pentagon to keep Anthropic's supply-chain-risk designation, preserving its ability to remove Claude from Defense Department systems and bar its use in department contracts. The D.C. Circuit ruled 2-1 against Anthropic in a dispute over restrictions on using the company's models for mass surveillance and autonomous weapons. Anthropic says the designation has cost it billions in business and damaged its standing ahead of a potential IPO. The decision is a direct adverse development for a provider seeking public-sector revenue: it leaves a large customer closed to Claude while the litigation continues, even though a separate California court has blocked a parallel government designation under another law.https://www.insighton.ai/anthropic/stories/plugins/https://www.insighton.ai/anthropic/stories/plugins/Fri, 25 Sep 2026 11:00:00 GMTAnthropic opened a portal on 25 September for developers on paid Claude plans to submit plugins to its directory, follow the review and see usage analytics after publication. A plugin can consist of a remote Model Context Protocol connector or a bundle combining MCP servers and Agent Skills; Claude Code bundles can also include language-server features, commands, hooks and agents. The portal automatically validates and safety-scans submissions, gives feedback, and lets approved developers choose when to publish. The move makes it easier for outside software makers to distribute capabilities inside Claude rather than ask each customer to configure a connector manually. Anthropic says millions of people already connect Claude to apps, work tools and data.https://www.insighton.ai/anthropic/stories/akamai/https://www.insighton.ai/anthropic/stories/akamai/Thu, 24 Sep 2026 11:00:00 GMTAnthropic committed about $11.6 billion over seven years for Akamai cloud services in a deal publicly announced on 24 September, with a warrant that could give the Claude developer as much as roughly 5% of Akamai's common equity on conversion and exercise. Akamai says the capacity will serve Anthropic's growing CPU workload, an important detail amid attention focused almost entirely on AI accelerators. The initial commitment is tied to about 2% of Akamai's shares through the warrant, while the remaining approximately 3% would vest if Anthropic expands the arrangement by another $9 billion. The agreement secures a substantial infrastructure option and links Anthropic's purchasing decisions to potential ownership in a supplier.https://www.insighton.ai/anthropic/stories/slack-connectors/https://www.insighton.ai/anthropic/stories/slack-connectors/Thu, 24 Sep 2026 11:00:00 GMTAnthropic began rolling out personal connectors for Claude Tag in Slack channels on 24 September, letting a person ask Claude to combine their own calendar, drive, CRM or deployment access with tools attached to a shared channel. Previously, the beta channel assistant could use only connectors an administrator attached to that channel. The feature is rolling out on Team plans, with Enterprise support to follow. It addresses a concrete problem in collaborative AI: much of the information needed to answer a team question belongs to one member and should not become a standing permission for everyone in the channel. Claude can now use that person's access for their request while keeping the channel's shared tools under their separate service identity.https://www.insighton.ai/anthropic/stories/dna-enzyme/https://www.insighton.ai/anthropic/stories/dna-enzyme/Wed, 23 Sep 2026 11:00:00 GMTAnthropic said on 23 September that Claude agents found an uncharacterized enzyme system in bacteriophage DNA, and human scientists at its new biology laboratory followed up with experiments. Roughly 950 agents spent 21 hours and 210 million tokens searching and analyzing sequence data before one noticed a repeating DNA pattern near a reverse-transcriptase gene. The team named the candidate array-associated reverse transcriptases, or ART, and published a preprint. The pattern resembles one feature of CRISPR systems, but the team has not established ART's biological function or a gene-editing use. The result is an early, concrete test of whether a general AI model can contribute a useful hypothesis to laboratory science, rather than merely summarize existing papers.