The LLM took an entirely different route, using a formula that was well known in related parts of math, but which no one had thought to apply to this type of question.

Liam Price just cracked a 60-year-old problem that world-class mathematicians have tried and failed to solve. He's 23 years old and has no advanced mathematics training.

I had the intuition that these problems were kind of clustered together and they had some kind of unifying feel to them. And this new method is really confirming that intuition.

The LLM took an entirely different route, using a formula that was well known in related parts of math, but which no one had thought to apply to this type of question.

This card was updated on April 24, 2026, to include additional information about safeguards for the deployment of GPT‑5.5 and GPT‑5.5 Pro in the API.

We separately evaluate GPT‑5.5 Pro in certain cases because we judge that the setting could materially impact the relevant risks or appropriate safeguards posture.

We are releasing GPT‑5.5 with our strongest set of safeguards to date, designed to reduce misuse while preserving legitimate, beneficial uses of advanced capabilities.

GPT‑5.5 understands the task earlier, asks for less guidance, uses tools more effectively, checks it work and keeps going until it's done.

We subjected the model to our full suite of predeployment safety evaluations and our Preparedness Framework, including targeted red-teaming for advanced cybersecurity and biology capabilities

Accepted applicants and collaborators must have existing ChatGPT accounts to apply

Once selected, successful applicants will be onboarded to the bio bug bounty platform

All prompts, completions, findings, and communications are covered by NDA

Smaller awards may be granted for partial wins at our discretion

$25,000 to the first true universal jailbreak to clear all five questions

Testing universal jailbreaks for biorisks in GPT‑5.5

We believe this is what drove the separate reports of usage limits draining faster than expected.

As part of this investigation, we ran more ablations (removing lines from the system prompt to understand the impact of each line) using a broader set of evaluations. One of these evaluations showed a 3% drop for both Opus 4.6 and 4.7.

After multiple weeks of internal testing and no regressions in the set of evaluations we ran, we felt confident about the change and shipped it alongside Opus 4.7 on April 16.

Two unrelated experiments made it challenging for us to reproduce the issue at first: an internal-only server-side experiment related to message queuing; and an orthogonal change in how we display thinking suppressed this bug in most CLI sessions

In our internal evals and testing, medium effort achieved slightly lower intelligence with significantly less latency for the majority of tasks.

We reverted this change on April 7 after users told us they'd prefer to default to higher intelligence and opt into lower effort for simple tasks.

The products will need to get worse, more expensive, or both if VCs are to get their money back.

the system achieved this training result more than 20 times faster than conventional synchronization methods.

chips from different generations running at different speeds still matched the ML performance of single-chip-type training runs, ensuring that even older hardware can meaningfully accelerate AI training.

With increasing levels of hardware failure, Decoupled DiLoCo continues to deliver a high level of 'goodput', or useful training, while that of other approaches nosedives.

By dividing large training runs across decoupled 'islands' of compute, with asynchronous data flowing between them, this architecture isolates local disruptions so that other parts of the system can keep learning efficiently.

Amazon is investing $5 billion in Anthropic today, with up to an additional $20 billion in the future. This builds on the $8 billion Amazon has previously invested.

Claude remains the only frontier AI model available to customers on all three of the world's largest cloud platforms: AWS (Bedrock), Google Cloud (Vertex AI), and Microsoft Azure (Foundry).

Anthropic will also use incremental capacity for Claude in Amazon Bedrock. The agreement includes expansion of inference in Asia and Europe to better serve Claude's growing international customer base.

Our run-rate revenue has now surpassed $30 billion, up from approximately $9 billion at the end of 2025.

We have signed a new agreement with Amazon that will deepen our existing partnership and secure up to 5 gigawatts (GW) of capacity for training and deploying Claude

The Prompt API uses the Gemini Nano model in Chrome. While the API is built into Chrome, the model is downloaded separately the first time an origin uses the API.

The Prompt API for the web is still being developed. While we build this API, refer to our best practices on session management for optimal performance.

The network requirement is only for the initial download of the model. Subsequent use of the model does not require a network connection. No data is sent to Google or any third party when using the model.

The Prompt API isn't available in Web Workers for now, due to the complexity of establishing a responsible document for each worker in order to check the permissions policy status.

Microsoft continues to participate directly in OpenAI's growth as a major shareholder.

Revenue share payments from OpenAI to Microsoft continue through 2030, independent of OpenAI's technology progress, at the same percentage but subject to a total cap.

Microsoft will continue to have a license to OpenAI IP for models and products through 2032. Microsoft's license will now be non-exclusive.

Microsoft will no longer pay a revenue share to OpenAI.

OpenAI can now serve all its products to customers across any cloud provider.

The compliance-driven buyers improvising local AI out of retail Mac Minis because the product they need does not exist.

Why the company that moved computing off the mainframe fifty years ago is making the same structural move with AI, and what that predicts.

The question it forces is not which model is best. It is who owns the inference layer your organization depends on, what happens when the economics of that layer stop being subsidized, and whether the thing in your pocket turns out to matter more than the thing in the datacenter.

The structural cost problem in AI inference that makes Apple's on-device bet defensible, not just defensive.

The board looked at the AI race Apple was losing and, rather than try harder at the thing that was failing, changed which game the company plays.

For a company that spent fifteen years running a functional model where no single discipline owned a product, putting two hardware engineers at the top is not a personnel decision. It is a structural break.

This means that improvements on SWE-bench Verified no longer reflect meaningful improvements in models' real-world software development abilities. Instead, they increasingly reflect how much the model was exposed to the benchmark at training time.

Tests reject correct solutions: We audited a 27.6% subset of the dataset that models often failed to solve and found that at least 59.4% of the audited problems have flawed test cases that reject functionally correct submissions

Our RL infra team used a K2.6-backed agent that operated autonomously for 5 days, managing monitoring, incident response, and system operations, demonstrating persistent context, multi-threaded task handling, and full-cycle execution from alert to resolution.

Kimi K2.6 autonomously overhauled exchange-core, an 8-year-old open-source financial matching engine. Over a 13-hour execution, the model iterated through 12 optimization strategies, initiating over 1,000 tool calls to precisely modify more than 4,000 lines of code.

Claude is now being deployed to NEC Group employees around the world

NEC will establish a Center of Excellence to develop a highly skilled, AI-enabled engineering organization

As part of its long-running Client Zero initiative, in which NEC serves as its own first customer before offering its technology to clients

NEC aims to build one of Japan's largest AI-native engineering teams, who will use Claude Code in their work.

Claude packages everything into a handoff bundle that you can pass to Claude Code with a single instruction.

Our most complex pages, which took 20+ prompts to recreate in other tools, only required 2 prompts in Claude Design.

What used to take a week of back-and-forth between briefs, mockups, and review rounds now happens in a single conversation.

Claude Design gives designers room to explore widely and everyone else a way to produce visual work.

Even experienced designers have to ration exploration—there's rarely time to prototype a dozen directions, so you limit yourself to a few.

The gains are especially strong in agentic coding, computer use, knowledge work, and early scientific research—areas where progress depends on reasoning across context and taking action over time.

GPT‑5.5 found a proof of a longstanding asymptotic fact about off-diagonal Ramsey numbers, later verified in Lean. The result is a concrete example of GPT‑5.5 contributing not just code or explanation, but a surprising and useful mathematical argument in a core research area.

We are treating the biological/chemical and cybersecurity capabilities of GPT‑5.5 as High under our Preparedness Framework. While GPT‑5.5 didn't reach Critical cybersecurity capability level, our evaluations and testing showed that its cybersecurity capabilities are a step up compared to GPT‑5.4.

Losing access to GPT‑5.5 feels like I've had a limb amputated.

GPT‑5.5 delivers this step up in intelligence without compromising on speed: larger, more capable models are often slower to serve, but GPT‑5.5 matches GPT‑5.4 per-token latency in real-world serving, while performing at a much higher level of intelligence.

GPT‑5.5 found a proof of a longstanding asymptotic fact about off-diagonal Ramsey numbers, later verified in Lean. The result is a concrete example of GPT‑5.5 contributing not just code or explanation, but a surprising and useful mathematical argument in a core research area.

The viable path is trusted access, robust safeguards that scale with capability, and the operational capacity to detect and respond to serious misuse.

We are treating the biological/chemical and cybersecurity capabilities of GPT‑5.5 as High under our Preparedness Framework. While GPT‑5.5 didn't reach Critical cybersecurity capability level, our evaluations and testing showed that its cybersecurity capabilities are a step up compared to GPT‑5.4.

GPT‑5.5 delivers this step up in intelligence without compromising on speed: larger, more capable models are often slower to serve, but GPT‑5.5 matches GPT‑5.4 per-token latency in real-world serving, while performing at a much higher level of intelligence.

The gains are especially strong in agentic coding, computer use, knowledge work, and early scientific research—areas where progress depends on reasoning across context and taking action over time.

We are treating the biological/chemical and cybersecurity capabilities of GPT‑5.5 as High under our Preparedness Framework. While GPT‑5.5 didn't reach Critical cybersecurity capability level, our evaluations and testing showed that its cybersecurity capabilities are a step up compared to GPT‑5.4.

GPT‑5.5 is priced higher than GPT‑5.4, it is both more intelligent and much more token efficient. In Codex, we have carefully tuned the experience so GPT‑5.5 delivers better results with fewer tokens than GPT‑5.4 for most users

The viable path is trusted access, robust safeguards that scale with capability, and the operational capacity to detect and respond to serious misuse.

GPT‑5.5 is our strongest agentic coding model to date. On **Terminal-Bench 2.0,** which tests complex command-line workflows requiring planning, iteration, and tool coordination, it achieves a state-of-the-art accuracy of 82.7%.

GPT‑5.5 delivers this step up in intelligence without compromising on speed: larger, more capable models are often slower to serve, but GPT‑5.5 matches GPT‑5.4 per-token latency in real-world serving, while performing at a much higher level of intelligence.

Jeremy didn't get laid off. He got leveraged.

A US lab would never; well, unless you count a code red or Meta's throw money at the problem moves.

The tradeoff is that the same input can map to more tokens—roughly 1.0–1.35× depending on the content type. Second, Opus 4.7 thinks more at higher effort levels, particularly on later turns in agentic settings. This improves its reliability on hard problems, but it does mean it produces more output tokens.

Our alignment assessment concluded that the model is 'largely well-aligned and trustworthy, though not fully ideal in its behavior'. Note that Mythos Preview remains the best-aligned model we've trained according to our evaluations.

On some measures, such as honesty and resistance to malicious 'prompt injection' attacks, Opus 4.7 is an improvement on Opus 4.6; in others (such as its tendency to give overly detailed harm-reduction advice on controlled substances), Opus 4.7 is modestly weaker.

Opus 4.7 is better at using file system-based memory. It remembers important notes across long, multi-session work, and uses them to move on to new tasks that, as a result, need less up-front context.

Interestingly, this means that prompts written for earlier models can sometimes now produce unexpected results: where previous models interpreted instructions loosely or skipped parts entirely, Opus 4.7 takes the instructions literally.

Several correlated but not strictly identical changes happened over the same few months: scaling inference compute, heavier use of RL in post-training, and models producing reasoning tokens.

Tasks where correctness is harder to verify may not have seen the same speedup, so the acceleration we document here may not be as general as the headline numbers suggest.

WeirdML V2 places models in an unusually resource-constrained environment: models get only five attempts to submit working code, with no access to external tools. This setup has not been the focus of recent RL training.

The three metrics where we find acceleration are concentrated in programming and mathematics. These are areas that labs have explicitly targeted for improvement, and they share an important property: correctness is easy to verify automatically.

Reasoning models show both a one-off jump in performance and a roughly 2-3x faster trend compared to non-reasoning models.

Three of four metrics show strong evidence of acceleration, seemingly driven by reasoning models.

Three of four metrics show strong evidence of acceleration, seemingly driven by reasoning models.

A free, good-enough product is enough to change market dynamics.

The commoditization flywheel : both companies give away complements to drive usage of the core.

Commoditizing complements doesn't always work because focus is scarce even for the largest, fastest growing businesses.

A free, good-enough product is enough to change market dynamics.

Some categories never developed a competitive response to this strategy : email, advertising infrastructure, user-generated video.

The risk of this strategy to the ecosystem is that it makes previously attractive categories no longer viable. Commoditizing the complement does not demand a best-in-class replacement.

If 90% is AI, do we even need a team?

Cursor counted the entire file as AI, even though we can see from the diff that it left plenty of the lines unchanged.

customers should expect PCW values of 85%+, often 95%+. This is not a hallucination and is accurate given how we compute this metric

placing constraints upon them not only helps users and services build trust in them, but it also helps people more easily conceptualise what they do.

lack of a well-defined user agent role in AI that's backed up by transparent, public standards... leaves a gap – it makes it harder for a marketplace to form.

Every time you use an Internet-connected computer, you're trusting someone (and most likely, a multitude) to act on your behalf.

The agent interprets new information and adapts the logic. The engine applies that logic continuously and emits precise updates.

Agents and CDC streams are powerful together because they split the work well.

With change data capture (CDC), the system emits a stream of precise updates: inserts, updates, deletes, each tied to specific records.

The fix is not smarter prompts. It is software built to meet agents halfway.

Humans are not a good target for calm technology.

Today's agents, the copilots, the chatbots are designed to be human like.

A LeadDev survey found 54% of engineering leaders believe AI copilots will reduce junior hiring long-term.

When juniors skip debugging and skip the formative mistakes, they don't build the tacit expertise. And when my generation of engineers retires, that knowledge doesn't transfer to the AI.

The Pentagon told defense CEOs to consolidate or die. Fifty-one major defense contractors collapsed into five.

A nuclear weapons program lost the ability to make a material it invented. The knowledge existed only in people, and the people were gone.

memory-driven experience scaling represents a crucial new frontier for agent scaling

simple checklists evolved into memories with compositional, preventative logic structures

existing TTS methods often discard the exploration trajectory and treat the final answer as the only useful outcome

this self-judgement does not need to be perfectly accurate, as we find ReasoningBank to be quite robust against judgment noise

by over-emphasizing successful experiences, they miss out on a primary source of learning — their own failures

by recording detailed actions instead of tactical foresight, they fail to distill higher-level, transferable reasoning patterns

When a Fugu model is allowed to call itself recursively, reading its own prior output as context and deciding whether to revise its coordination strategy, a new form of test-time scaling emerges.

A core conviction at Sakana AI is that the most capable AI systems will not be monolithic models scaled in isolation, but collections of specialized agents working together.

Sakana Fugu coordinates pools of frontier foundation models to achieve state-of-the-art performance across coding, mathematics, scientific reasoning, etc.

Chinese authorities have reportedly been pushing data centers and public computing projects to use more domestic chips, including through reported bans on foreign-made chips, sourcing quotas, and requirements to pair Nvidia chips with Chinese alternatives.

DeepSeek does not appear to have fully moved beyond Nvidia. The company's technical report reveals that it is using Chinese chips to run the model for inference, but...appears to have adapted only part of V4's training process for Chinese chips.

DeepSeek V4 exceeds them all on coding, math, and STEM problems, making it one of the strongest open-source models ever released.

The filing cabinet keeps getting bigger. But a bigger filing cabinet is still a filing cabinet.

The current separation between training and deployment is not just an engineering convenience – it is a safety, auditability, and governance boundary.

The intelligence lives in the static parameters, and the apparent capabilities change radically depending on what you feed into the window.

Without our safeguards in place (which we do to measure a model's raw capabilities), only Mythos Preview and Opus 4.7 completed more than half the tasks.

We also welcome feedback and input from third parties and industry experts. We're currently working with The Future of Free Speech (an independent think tank at Vanderbilt University), the Foundation for American Innovation, and the Collective Intelligence Project

if AI models can answer these questions well (that is, accurately and impartially), they can be a positive force for the democratic process.

No encryption protects that layer. The router can read, change, or replace anything.

Out of 28 paid and 400 free routers: > 9 injected malicious code into tool calls > 17 touched researcher-owned AWS credentials > 1 drained $500k from an Ethereum wallet

Client-side defenses caught 89% of injections. But real protection needs providers to sign responses.

Some attacks only fired after 50 prior calls. Others activated only in auto-approve mode.

a lightweight surrogate trained on them can absorb a significant portion of future traffic at near-zero marginal inference cost

On a 150-class benchmark, the surrogate fully replaces the teacher

Duplicate section heading | Error; reject the file

Unknown component property | Accept with warning

The DESIGN.md format is at version `alpha`. The spec, token schema, and CLI are under active development. Expect changes to the format as it matures.

Components map a name to a group of sub-token properties: ... Variants (hover, active, pressed) are expressed as separate component entries with a related key name.

Unknown section heading | Preserve; do not error

A DESIGN.md file combines machine-readable design tokens (YAML front matter) with human-readable design rationale (markdown prose). Tokens give agents exact values. Prose tells them _why_ those values exist and how to apply them.

Unknown component property | Accept with warning

Components map a name to a group of sub-token properties: ... Valid component properties: backgroundColor, textColor, typography, rounded, padding, size, height, width.

A DESIGN.md file combines machine-readable design tokens (YAML front matter) with human-readable design rationale (markdown prose). Tokens give agents exact values. Prose tells them _why_ those values exist and how to apply them.

At our request, the underwriters have reserved up to _______% of the shares of Class A common stock offered by this prospectus for sale at the initial public offering price through a directed share program to certain persons identified by our management and certain long-tenured employees, which may include parties with whom we have a business relationship and friends and family of management and such employees.

We have applied to list our Class A common stock on the Nasdaq Global Select Market under the symbol 'CBRS,' and this offering is contingent upon the listing of our Class A common stock on the Nasdaq Global Select Market.

Distributed systems background - real-time sessions, cloud infrastructure (AWS), and production reliability

Hands-on experience building or maintaining a game engine (must), with strong Python fundamentals (must)

Help lay the game and environment foundations for ARC-AGI-4 and ARC-AGI-5

You'll be responsible for stabilizing the current stack to setting the foundation for what comes next.

A senior engineer to own and evolve the game engine and real-time play infrastructure behind the ARC-AGI series.

Begin with fast checks that explain behavior, then add slower, heavier checks only when they reduce risk.

The fastest way to get started is to use Codex's built-in skill creator (which itself is also a skill).

The most reliable way to improve a skill over time is to evaluate it the same way you would any other prompt for LLM applications.

It happens several times a year in the US alone, often unreported, and about 100 times a year worldwide.

Nor does it matter, given that the modifying strains for pathogens for research purposes is what every research lab does, because that is what virology is.

For me Ralph Baric's 2024 test testimony moved the lab leak hypothesis to pretty likely.

And since then, there is no more scientific evidence or verifiable sources. Hence the reason the CIA didn't even believe it and gave it the lowest confidence rating it has.

We run a similar model processing loan documents that would normally require a team of 15.

This is the part people miss about AI-native companies - the $113k is not a cost, it is your headcount budget allocated differently.

Our goal is $10M ARR [annual recurring revenue] with a sub-10 person org.

The real unlock is compound scaling—token spend grows linearly while output grows exponentially.

We imagine a world where all of the tools you use are as rich and visual as the world we live in.

If the most effective way to communicate something were a single word, an illustration, or a photorealistic rendering, that's what you'd see.

The screen you're reading this on is already presenting you an image, it's just generated with rigid code and rules that makes it difficult to communicate complex and detailed ideas.

All text on the screen is rendered as pixels by the image model. There are no text overlays applied to the images.

The entire web is just generated pixels on your screen.

Vercel is advising Google Workspace administrators and Google account owners to check for the following application: OAuth App: 110671459871-30f1spbu0hptbs60cb4vsmv79i7bbvqj.apps.googleusercontent.com

Unfortunately, the attacker got further access through their enumeration.

Vercel stores all customer environment variables fully encrypted at rest. We have numerous defense-in-depth mechanisms to protect core systems and customer data.

The action that matters most — building semiconductor-grade hydrogen bromide gas conversion capacity outside Israel — takes years.

The structural failure is not the war: It is that the global memory supply chain has built itself around a conversion chokepoint with no redundancy and no fallback.

The story receiving almost no attention is bromine, and it is potentially the more dangerous one.

Emissions in advanced economies grew faster (+0.5%) than in emerging and developing economies (+0.3%) for the first time since the 1990s.

Battery storage was the fastest-growing power technology, with around 110 gigawatts (GW) of new capacity added – more than any year of natural gas capacity additions on record.

Solar was the single biggest contributor to global energy supply growth in 2025. It accounted for more than 25% of the increase – the first time a modern renewable has led global primary energy growth.

we probably will publish more curl vulnerabilities in 2026 than we have done in many years, maybe ever.

it is decently important to handle them asap when they arrive so that we can avoid building up too much backlog.

The time when we suffer from large amounts of AI slop is gone. Now we instead suffer under a massive load of good reports.

In addition to empowering developers and agents to handle project setup and boilerplate code, we've also designed these new tools and resources to make it easier to transition to Android Studio.

By accessing the frequently updated knowledge base, agents can ground their responses in the most recent information from Android developer docs, Firebase, Google Developers, and Kotlin docs. This ensures that even if an LLM's training cutoff is a year old, it can still provide guidance on the latest frameworks and patterns we recommend today.

Android skills cover some of the most common workflows that some Android developers and LLMs may struggle with—they help models better understand and execute specific patterns that follow our best practices and guidance on Android development.

The new Android CLI serves as the primary interface for Android development from the terminal, featuring commands for environment setup, project creation, and device management—with more modern capabilities and easy updatability in mind.

Whether you are using Gemini in Android Studio, Gemini CLI, Antigravity, or third-party agents like Claude Code or Codex, our mission is to ensure that high-quality Android development is possible everywhere.

In our internal experiments, Android CLI improved project and environment setup by reducing LLM token usage by more than 70%, and tasks were completed 3X faster than when agents attempted to navigate these tasks using only the standard toolsets.

Capture sequences and replay them as stable automations.

Legacy workflows, turned into dependable APIs for agents and internal systems.

Map tokens, cookies, sessions, and multi-step sequences automatically.

See every HTTP/S request from any app or browser in real time.

Maintains your HTTP/TLS fingerprint so intercepted traffic behaves identically to the original.

Kampala lets you reverse engineer anything including websites, mobile apps, and desktop apps instantly.

SWE-chat is a living dataset; our collection pipeline automatically and continually discovers and processes sessions from public repositories

despite rapidly improving capabilities, coding agents remain inefficient in natural settings

users push back against agent outputs -- through corrections, failure reports, and interruptions -- in 44% of all turns

agent-written code introduces more security vulnerabilities than code authored by humans

Just 44% of all agent-produced code survives into user commits

coding patterns are bimodal: in 41% of sessions, agents author virtually all committed code ('vibe coding'), while in 23%, humans write all code themselves.

The overall conclusion, therefore, is that AI for Science should be understood as both a scientific and a civilizational project.

The central question is not whether AI can imitate human conversation, but whether it can participate in the production of publishable scientific knowledge at a level comparable to a recognized human contributor.

Without a mechanism for continuous and diverse learning, AI systems will tend to reproduce the dominant patterns already present in their training data. That limitation would make truly creative work difficult.

The most effective pattern of human-AI cooperation may differ substantially across disciplines, and these patterns will likely be discovered through practice rather than designed in advance.