A Tencent Research Institute essay republished by 36Kr argues that humanity is entering a dual-intelligence era in which human and AI cognition coexist as distinct but interacting systems. It cites research on AI-assisted Go play and examples of AI-assisted scientific discovery to examine how AI may expand human problem-solving while introducing new governance questions. The essay proposes “human-centered intelligence,” a coexistence framework preserving human rights to set goals, assess developments, and stop or correct AI systems.
The Decoder reports that Anthropic's Fable 5.2 completed 41.3% of real freelance projects at professional quality in Scale Labs and the Center for AI Safety's Remote Labor Index. The cited score was about twice that of Fable 5.1 and ahead of Opus 5.5 and GPT 6 Astra, while nearly 50 successful projects had not been solved by any tested agent.
The Frontier Model Forum released version 1.0 of its Jailbreak Rating Framework, a proposal modeled loosely on CVSS that rates LLM jailbreaks by capability gain, capability scope, deployment ease, and discovery ease. Tom’s Hardware reports that the framework combines severity and likelihood into a rating from low to critical, with the capability-gain score constraining the other dimensions. FMF says the cross-vendor scale could help teams developing patches, but version 1.0 remains under review, and the group advises against relying on it while comments and suggestions are collected through December 21.
Bloomberg Technology reports that Anthropic is seeking a fully diluted valuation of about $1.4 trillion in its initial public offering. The Claude maker aims to become the first frontier AI lab to go public and could raise as much as $90.4 billion through the IPO and a private placement.
A Reddit user says Artificial Analysis ranks Fable 5.2 above Claude Opus 5.5, with performance rising smoothly from High to XHigh and jumping more noticeably at Max. The author recommends Opus 5.5 XHigh, Fable 5.2 High, or Fable 5.2 Max, describing Fable 5.2 High as the preferred switching point.
36Kr reports that AI companies are turning agent runtime harnesses into hosted services. Microsoft expanded Copilot Harness across its products and offered managed services through the cloud, while Moonshot AI released Kimi Hosted Agents with a “Harness as a Service” offering. OpenAI and Anthropic had also made their internal runtime systems available to developers, while DeepSeek open-sourced its harness framework.
In a Microsoft technical white paper, Microsoft reported that keeping Claude Opus 4.8 unchanged but switching the harness behind Copilot Studio increased the complex-task pass rate by 22.1 percentage points. The new system nearly doubled model calls per task, yet reduced calculated input-token consumption by 21%.
Anthropic has reportedly begun an IPO roadshow targeting a valuation of $1 trillion. The supplied report provides no additional details or independent confirmation of the target valuation.
Anthropic set a tentative valuation of $1.26 trillion to $1.4 trillion for its initial public offering, according to an updated securities filing. The company seeks to raise $90 billion. The valuation range is slightly below the $1.5 trillion at which Anthropic was expected to go public just last month.
Sources say Anthropic set an initial public offering price range of $70 to $78 per share and is expected to begin trading on Nov. 19. The IPO would come roughly five months after Elon Musk took SpaceX public in a record offering.
A 36Kr flash citing CLS reports that Anthropic expects an IPO price midpoint of $74 per share. It says net proceeds would reach $93.7 billion after overallotment and private placements.
TSMC's board approved a capital budget of about US$40.927 billion covering advanced-process installations and upgrades, advanced packaging, mature and specialty capacity, fab construction, 2027 R&D, maintenance spending, and capitalized leases. TechNews also reported that the board approved strategic equity purchases of up to US$500 million each in Anthropic and OpenAI, plus the establishment of an investment company with up to US$500 million in funding. TSMC also approved a third-quarter 2026 cash dividend of NT$9 per share alongside reported quarterly revenue of NT$1.49424 trillion and EPS of NT$29.43.
A V2EX user reported that NUS faculty and staff can now purchase Claude for Education subscriptions through official Claude channels. Unlike the university’s earlier web-based aggregation, the new option provides direct official-channel accounts; the report says the earlier service did not support API or Claude Code use. The standard plan is listed at SGD 280 per year and the advanced plan at SGD 790 per year, with both covering all current Claude models. The user’s comparisons to rumored Max 2.5x and 12.5x usage tiers remain unverified, and whether five-hour limits apply is unclear.
Anthropic files reportedly showed that compute and infrastructure expenses reached $9.9 billion in Q3 2026, equal to 58% of quarterly revenue. The share rose from 48% in Q2 and included $5 billion in short-term compute commitments that were substantially more expensive than comparable capacity under multi-year agreements.
Anthropic said its annualized revenue reached about $81 billion as of October 31, 2026, up $10 billion from September 30. The company also said the number of customers paying more than $100,000 annually had risen to 11,900, according to 36Kr and CLS.
A 36Kr brief attributed to Sina Finance reported that Anthropic’s target valuation for its IPO is approximately $1 trillion. The item did not identify Anthropic or a representative as the source, provide supporting details, or explain the basis for the target.
Anthropic said it had launched the roadshow for its proposed initial public offering. The company said a registration statement for the securities had been filed with the SEC but had not yet taken effect. The securities may not be sold, and offers to buy may not be accepted, before the registration statement becomes effective.
Anthropic launched its IPO roadshow, proposing 825 million Class A shares at $70–$78 each, with 801.35 million offered by the company and about 23.65 million sold by existing shareholders. It also filed for a separate $26 billion private placement involving Google, Nvidia, AMD, Broadcom, Micron, Microsoft, and TSMC Partners. The Next Web reported that the offering was oversubscribed, while Financial Times reporting indicated bankers were considering a valuation of $1.5 trillion or more; final pricing and the listing date remained unconfirmed.
A report says Edward Witten credited Claude Opus 5.5 with proposing a possible mathematical construction and helping with technical details in an arXiv preprint on lattice definitions of gauge theories with chiral fermions. In related preprints, Nathaniel Craig credited GPT-6 Astra with substantial mathematical derivations and Lean-based verification, while Stephen D. H. Hsu said GPT and Claude helped produce technical details and check results. All three papers remain preprints and require further examination by researchers.
Anthropic released Claude Fable 5.2 and Mythos 5.2, describing both as cheaper and faster upgrades, with Fable 5.2 priced at $8 per million input tokens and $40 per million output tokens. Artificial Analysis placed Fable 5.2 first on its Intelligence Index, five points ahead of the unreleased Gemini 4 Carbon, while Mythos 5.2 is limited to vetted security and life-sciences groups. Anthropic says Fable 5.2 is 20% cheaper and up to 30% faster than version 5.1, though it remains twice as expensive as Opus 5.5.
A social-media post attributes an interactive ink-themed project to Fable 5.2. The user says they asked it to create a project demonstrating its abilities without specifying what to build. According to the post, typed characters generate corresponding ink imagery, the ink uses a real-time fluid simulation, the mouse controls where each character appears, and the project includes sound, music, dynamic objects and interacting weather effects.
A DoNews headline reports that Anthropic acknowledged sending incorrect documents and would update them. It also mentions “26 billion anchoring” and a “69 billion plan,” but does not specify the currency or provide enough context to establish what those figures concern.
Claude Code initially added a rule prohibiting users from reproducing an existing game, even if they claimed the result was original, and its developers confirmed the change. After developers said the restriction disrupted legitimate work, the rule was relaxed within a week: fully copying a game remains restricted, while developing similar games by referencing gameplay mechanics is again permitted. The report also documents independent studios hiding demos amid AI-assisted copying, alongside reactive platform enforcement and unresolved legal boundaries.
A Juejin AI article relays Anthropic Engineering’s Contextual Retrieval method, which adds document-level explanations to chunks after splitting and before embedding or keyword indexing. Anthropic reports that Contextual Embeddings combined with Contextual BM25 reduced retrieval failures by 49%, while adding reranking reduced failures by up to 67% in its tests. These benchmark figures are vendor claims relayed by a secondary source.
Anthropic reportedly sent banks IPO documents outlining about 866.7 million Class A shares at $70–$80 each, a maximum transaction size of roughly $69.3 billion, then withdrew the pricing details as inaccurate. The company did not explain the error or comment, while Bloomberg described the episode as a brief clerical mistake; final pricing and fundraising totals remain unsettled. The documents also showed that Alphabet, Nvidia, AMD, Broadcom, Micron, Microsoft and TSMC planned to buy $26 billion of non-voting Class N shares at the IPO price.
QbitAI gathered user demonstrations of Fable 5.2 spanning game development, 3D modeling, video generation, music composition and agent workflows. The report says benchmarks showed the model nearly matching or exceeding Opus 5.5, with output up to 30% faster than Fable 5.1, support for a 1-million-token context and up to 128,000 output tokens, and API prices 20% lower. User examples included playable games, a piano sonata and visualizer, code-generated videos, a Blender scene and hundreds of agent-generated 3D worlds. The report also noted complaints about rapid token consumption at higher reasoning settings and some tasks where gains over Opus 5.5 were limited.
A supplied report headline says Anthropic released Claude Fable 5.2 and Mythos 5.2, with improvements in performance and efficiency. The headline does not attribute the improvement claim, and the supplied report includes no technical details or benchmark results.
Anthropic is set to begin formally marketing its initial public offering as early as Tuesday, Bloomberg reported. The report said an apparent problem with filing documents created last-minute drama ahead of the marketing effort but did not specify the issue.
Anthropic’s documentation says Claude Fable 5.2 responds closely to code-ownership wording when working on a codebase. In Anthropic’s testing on small coding tasks, the model merged repeated code in about half the attempts when prompted that the codebase was its own, compared with fewer than one in ten attempts without that wording. Simon Willison’s Weblog quoted these findings from Anthropic’s “Prompting Claude Fable 5.2” documentation.
Bloomberg reported, citing people familiar with the matter, that Anthropic's IPO attracted orders several times the number of shares available. 36Kr and cnBeta also reported that CEO Dario Amodei and JPMorgan Chase CEO Jamie Dimon planned a Friday livestream about the offering. According to cnBeta, previously filed documents proposed selling 825 million shares at $70 to $78 each alongside a $26 billion private placement, for total proceeds of up to about $90.4 billion. The demand figures came from sources familiar with the matter, while the livestream and final offering terms were not confirmed as completed.
A V2EX user reported subscribing to Claude through the US App Store and said one subscription would permanently bind to the Claude account. The user also described being charged without receiving access, said a refund could be requested, and reported that switching accounts and retrying triggered a KYC prompt. Another user asked whether a new Claude account could be paired with a different Apple account. The post does not establish whether these subscription mechanics are generally available or how the account-binding rule works.
What problem it solves
Pi is designed for developers who need to adapt an agent to their own workflows rather than work within a fixed feature set. It provides interactive, automated, RPC, and SDK interfaces while leaving optional capabilities such as sub-agents and plan mode to extensions or packages.
Key features
Extensible agent harness — Customize the harness with extensions, skills, prompt te…
What problem it solves
Managing multiple AI service sessions and translating them into compatible API interfaces can require repetitive account switching, custom routing, and protocol-specific integration. The project addresses these tasks with a desktop application that combines account administration, protocol conversion, request scheduling, and model routing.
Key features
Account dashboard and …
What problem it solves
Managing subscriptions across multiple AI providers can require separate access paths and fragmented quota handling. This project provides a centralized gateway for distributing subscription quotas, supporting shared access and compatible use with native AI tools.
Key features
Subscription quota distribution — The gateway is designed to distribute and manage AI subscription …
What problem it solves
Pull request changes can contain security vulnerabilities that require understanding their code context and intent. The action helps identify potential issues, filter false positives, and communicate findings directly in pull requests.
Key features
Context-aware security analysis — Claude examines changed code in context to understand its purpose and potential security impli…
What problem it solves
Serving large language models requires efficient memory management, flexible model execution, and practical integration with existing models and APIs. vLLM addresses these needs with inference and serving capabilities spanning batching, caching, quantization, distributed execution, and multiple hardware platforms.
Key features
Efficient memory and serving — vLLM manages atte…
What problem it solves
AI agents used in security work can write code and search the web, but they lack the practitioner playbooks that turn a generic model into a capable security analyst. Existing security tool repositories offer wordlists, payloads, or exploit code, but not the structured decision-making workflow a senior analyst follows: when to use each technique, what prerequisites to check,…
What problem it solves
AI coding tools often expect particular provider protocols, making it difficult to use the models and providers a user prefers. A local gateway can translate between supported client APIs and provider APIs while preserving capabilities such as streaming, tool calls, reasoning, and images.
Key features
Provider protocol translation — The gateway translates Responses, Chat Com…
What problem it solves
AI agents often send large tool outputs, logs, files, RAG chunks, and conversation history to language models, increasing the input context that must be processed. Developers need to reduce that input while preserving important content, keeping compression local, and retaining access to original material when necessary.
Key features
Content-aware compression — A ContentRoute…
What problem it solves
The hidden system instructions that shape AI product behavior are difficult to inspect and change across models, products, and versions. This repository makes collected prompts available in a structured, product-oriented archive for study and comparison.
Key features
Broad product coverage — The archive covers prompts from major AI providers and products, including Anthropic…
What problem it solves
Building AI agents and workflows often requires combining visual orchestration with custom code while connecting them to existing systems. The platform addresses these needs with model flexibility, self-hosted or cloud deployment, workflow logic, human approvals, and observability.
Key features
Visual and code-based workflow building — Build workflows on a visual canvas and …
What problem it solves
Serving language models requires reading checkpoints, tokenizing requests, and running accelerated kernels while managing a local model server. TensorFold addresses these requirements with a native binary that can run on Apple Silicon and NVIDIA GPUs without requiring Python or MLX for serving.
Key features
Native Metal and CUDA inference — The Zig-based engine executes mode…
What problem it solves
Agent workflows need to use hosted, BYOK, and local models while applying access, usage, and spending controls. They also need a way to turn production traffic into routing decisions or a specialized model tailored to observed workloads.
Key features
Unified model access — Route hosted, BYOK, and local models through one OpenAI-compatible API, with support for connecting pro…
What problem it solves
AI agents need a reliable way to perform browser tasks such as finding availability, selecting options, handling CAPTCHAs, and completing forms. Developers also need flexibility to run the agent locally or through hosted infrastructure while choosing their model, browser mode, tools, and output format.
Key features
Persistent Python agent runtime — The model writes Python in…
What problem it solves
Personal AI agents need to connect language model providers, communication channels, and tools while remaining under the user's control. They also need configurable autonomy, security boundaries, and operational features such as background services, approvals, and persistent workflows.
Key features
Multiple channels and providers — The agent can receive messages through conf…
What problem it solves
Developers need a practical starting point for applications built with the Claude API. The quickstarts address this by providing foundations that can be customized for different agent, browser, computer-use, communication, and data-analysis needs.
Key features
Application quickstarts — The collection includes foundations for customer support, financial analysis, coding, and …
An Agent can complete a task once by relying on the context and reasoning available at that moment. To avoid relearning lessons on the next task, it needs to turn past experience into reusable knowledge. The challenge is that experience is never an isolated action: it also includes the original goal, environment, and why the action worked. Remembering only “how we did it last time” may save redund…
What problem it solves
Developers need a way to understand their codebase, carry out routine coding tasks, explain complex code, and handle git workflows without navigating multiple command-line tools for each task. Claude Code provides these capabilities through natural language commands in the terminal, an IDE, or GitHub.
Key features
Natural-language coding assistance — Claude Code can execute …
What problem it solves
Building, deploying, and managing AI agents requires integrating agent behavior, runtime configuration, tools, models, sessions, and observability. The project addresses this with a declarative Kubernetes-based framework that supports multiple runtimes, model providers, and MCP tools.
Key features
Declarative Agent Management — Agents are Kubernetes custom resources that com…
What problem it solves
Running frontier-scale open-weight models locally requires an inference platform that can make efficient use of the heterogeneous hardware available on a personal device. The project addresses this by treating GPUs, CPUs, host memory, and interconnects as a unified, elastic inference platform.
Key features
Edge-native MoE runtime — The runtime provides MoE serving with bandw…
What problem it solves
Configuring Claude Code can require assembling many kinds of reusable components and integrations. The project addresses this with a searchable component catalog, one-command installation, local monitoring tools, and installation diagnostics.
Key features
Searchable component catalog — Browse available skills, agents, commands, MCP integrations, settings, hooks, mods, loops,…