Although it dates back to the early days of the Marconi Company in the 1920s, the Franklin oscillator has remained a relatively obscure circuit, its memory mostly kept alive by ham radio operators who prize its high stability at higher frequencies. At the core of the circuit is an LC tank circuit, a fact which [nobcha] used to build quite a precise LC meter.

The meter is built around two parts: the Franklin oscillator, which resonates at a frequency defined by its inductance and capacitance, and an Arduino which counts the frequency of the signal. In operation, the Arduino measures the frequency of the original LC circuit, then measures again after another element (capacitor or inductor) has been added to the circuit. By measuring how much the resonant frequency changes, it’s possible to determine the value of the new element.

Before operation, the meter must be calibrated with a known reference capacitor to determine the values of the base LC circuit. In one iteration of the design, this was done automatically using a relay, while in a later version a manual switch connects the reference capacitor. Because the meter measures frequency differences and not absolute values, it minimizes parasitic effects. In testing, it was capable of measuring inductances as low as 0.1 µH.

We’ve seen a few homebrew LC meters here, some battery-powered and some rather professional.

Major wireless carriers: A necessary evil if you travel a lot, have a family, or are just interested in coverage that’s reliably consistent and widespread. AT&T is the third-largest provider in the US (first for 5G), with the largest coverage map. I’ve had various AT&T plans for more than a decade, first for just myself and now for my whole family, even though I only get one cell bar at my house and have to stand in one 5-square-foot patch of yard to make a phone call. And have lost entire days of my life to fighting unexpected random charges and upsells. (Verizon is somehow worse.) But anyway! AT&T is fine, it has all the latest phones, and there are some legitimately good perks, like no roaming in Canada or Mexico with select plans. If you know you’re going to have to go with one of the big guys, don’t sign up without checking out the below discounts first.

Save on AT&T Prepaid Phone Plans With the Latest Deals

An AT&T prepaid phone plan is one of the easiest ways to save big on your future phone bills. AT&T has a wide selection of prepaid phone plans, including 5G prepaid plans and multi-month long-term plans. For as low as $25 per month, you’ll get unlimited talk, text, and data. Plus, all AT&T prepaid plans include AT&T ActiveArmor mobile security, and are eligible for an eSIM or SIM card for as little as $0.99.

Get the new Samsung Galaxy S26 Ultra for $0

We on the WIRED Reviews team love the Samsung Galaxy S26 Ultra. We rated it a high 8/10 because of its built-in privacy display. We also loved the horizon lock to capture super steady video footage. Plus, it has excellent performance, great battery life, and a reliable quad-camera system. And right now, you can get a Samsung Galaxy S26 Ultra for free with an eligible trade-in (in any condition, but has a required trade-in of Galaxy S24+, Z Fold5, or newer).

Save Over $600 a Year With AT&T Fiber

AT&T Fiber claims to be the fastest internet network in America. You can find out for yourself (for less) with this new deal. When first time customers sign up for Fiber now, they’ll get 1 Gigabyte for only $37 per month. That’s over $600 in savings per year!

Are There AT&T Promos for Existing Customers?

But I already have AT&T, you might be saying—new deals never apply to me. Do you have AT&T internet, though? If, like me, you have AT&T for your phone plan and Xfinity or CenturyLink for internet, did you know you can save 20% off your AT&T bill every single month if you bundle your internet service with unlimited wireless? This applies to both current phone customers and current internet customers who don’t have both plans.

AT&T wants to reward you for your loyalty: when you sign up for AT&T Fiber and eligible wireless plans, you can get up to $150 in AT&T Visa Reward Cards. Be sure to check out the AT&T deals page for more details on that offer, along with other great ways to save.

AT&T Discounts for Professionals and Students

One of the reasons I went with AT&T in the first place is because my husband is a college professor and we get a generous 25% off our bill. This 25% discount applies for all teachers, as well as active military, veterans, first responders, and many medical professionals. Not one of the above? Students and union members can save $10 per line per month, and union members can also score 20% off accessories like cases, and stuff you used to be able to get with your phones, like chargers and cords.

How Can I Make My AT&T Plan Cheaper?

Not a new customer; not in a place to bundle; and not a teacher/first responder, in the military, or a student? All is not lost on the discount front. You can save over $800 a year on AT&T Wireless when you bundle four unlimited wireless plans with your current internet plan. (Savings based on 20% discount on four voice lines with eligible internet service, plus $10/month discount with eligible AutoPay & paperless bill, which starts within two bills.)

Don’t need four unlimited wireless plans? Check out if your employer offers a discount—a couple of mine have in the past, and you can save $10 per month per line on the unlimited plan. Check here to see if your workplace qualifies.

You can also get a discount on a new phone with an eligible trade-in, but the best deal yet may be the fact AT&T lets you try its wireless free for 30 days. Keep your current service and phone number while trying out AT&T’s network from your device—no catches or commitments. You don’t even need a credit card. It’s a great way to see if you get good service where you’ll be using the phone most.

Save More With AT&T Family Plans

If you want multiple people on your phone line or are adding a line for your child’s first phone, an AT&T Family Plan is one of the most cost effective ways to make the change. With AT&T family plans, you can mix and match any of AT&T’s unlimited plans to get great deals and serious discounts on any smartphone for each family member. Depending on what you choose, plans start at only $36 per month, per line (for 4 lines).

Choose AT&T for the Best Internet for Gaming

If you’re a big gamer, you’re going to want fast, reliable internet that’ll provide clear, bright graphics without laggy gameplay or interactive audio. AT&T Fiber with All-Fi promises to give gamers everything they look for in a service, with super fast speeds (up to 5 GIG), and tons of bandwidth for fast upload and download speeds.

The generative AI era began for most people with the launch of OpenAI’s ChatGPT in late 2022, but the underlying technology — the “Transformer” neural network architecture that allows AI models to weigh the importance of different words in a sentence (or pixels in an image) differently and train on information in parallel — dates back to Google’s seminal 2017 paper “Attention Is All You Need.”

Yet while Transformers deliver unparalleled model quality and have underpinned most of the major generative AI models used today, they are computationally gluttonous. They are burdened by quadratic compute and linear memory demands that make large-scale inference an expensive, often prohibitive, endeavor. Hence, the desire by some researchers to improve on them by developing a new architecture, Mamba, in 2023, which has gone on to be included in hybrid Mamba-Transformer models like Nvidia’s Nemotron 3 Super.

Now, the same researchers behind the original Mamba architecture including leaders Albert Gu of Carnegie Mellon and Tri Dao of Princeton have released the latest version of their new architecture, Mamba-3, as a language model under a permissive Apache 2.0 open source license — making it immediately available to developers, including enterprises for commercial purposes. A technical paper has also been published on arXiv.org.

This model signals a paradigm shift from training efficiency to an “inference-first” design. As Gu noted in the official announcement, while Mamba-2 focused on breaking pretraining bottlenecks, Mamba-3 aims to solve the “cold GPU” problem: the reality that during decoding, modern hardware often remains idle, waiting for memory movement rather than performing computation.

Perplexity (no, not the company) and the newfound efficiency of Mamba 3

Mamba, including Mamba 3, is a type of State Space Model (SSM).

These are effectively a high-speed “summary machine” for AI. While many popular models (like the ones behind ChatGPT) have to re-examine every single word they’ve already seen to understand what comes next—which gets slower and more expensive the longer the conversation lasts—an SSM maintains a compact, ever-changing internal state. This state is essentially a digital “mental snapshot” of the entire history of the data.

As new information flows in, the model simply updates this snapshot instead of re-reading everything from the beginning. This allows the AI to process massive amounts of information, like entire libraries of books or long strands of DNA, with incredible speed and much lower memory requirements.

To appreciate the leap Mamba-3 represents, one must first understand perplexity, the primary metric used in the research to measure model quality.

In the context of language modeling, perplexity is a measure of how “surprised” a model is by new data.

Think of a model as a professional gambler. If a model has high perplexity, it is unsure where to place its bets; it sees many possible next words as equally likely.

A lower perplexity score indicates that the model is more “certain”—it has a better grasp of the underlying patterns of human language. For AI builders, perplexity serves as a high-fidelity proxy for intelligence.

The breakthrough reported in the Mamba-3 research is that it achieves comparable perplexity to its predecessor, Mamba-2, while using only half the state size. This means a model can be just as smart while being twice as efficient to run.

A new philosophy

The philosophy guiding Mamba-3 is a fundamental shift in how we think about AI “intelligence” versus the speed of the hardware it runs on. While the previous generation, Mamba-2, was designed to be trained at record-breaking speeds, Mamba-3 is an “inference-first” architecture — inference referring to the way AI models are served to end users, through websites like ChatGPT or Google Gemini, or through application programming interfaces (APIs).

Mamba 3’s primary goal is to maximize every second the computer chip (GPU) is active, ensuring that the model is thinking as hard as possible without making the user wait for an answer.

In the world of language models, every point of accuracy is hard-won. At the 1.5-billion-parameter scale, the most advanced “MIMO” variant of Mamba-3 achieved a 57.6% average accuracy across benchmarks, representing a 2.2-percentage-point leap over the industry-standard Transformer.

While a two-point jump might sound modest, it actually represents a nearly 4% relative increase in language modeling capability compared to the Transformer baseline. Even more impressively, as alluded to above, Mamba-3 can match the predictive quality of its predecessor while using only half the internal “state size,” effectively delivering the same level of intelligence with significantly less memory lag.

For years, efficient alternatives to Transformers suffered from a “logic gap”—they often failed at simple reasoning tasks, like keeping track of patterns or solving basic arithmetic, because their internal math was too rigid. Mamba-3 solves this by introducing complex-valued states.

This mathematical upgrade acts like an internal compass, allowing the model to represent “rotational” logic. By using this “rotary” approach, Mamba-3 can near-perfectly solve logic puzzles and state-tracking tasks that its predecessors could only guess at, finally bringing the reasoning power of linear models on par with the most advanced systems.

The final piece of the puzzle is how Mamba-3 interacts with physical hardware. Most AI models today are “memory-bound,” meaning the computer chip spends most of its time idle, waiting for data to move from memory to the processor.

Mamba-3 introduces a Multi-Input, Multi-Output (MIMO) formulation that fundamentally changes this dynamic. By performing up to four times more mathematical operations in parallel during each step, Mamba-3 utilizes that previously “idle” power. This allows the model to do significantly more “thinking” for every word it generates without increasing the actual time a user spends waiting for a response. More on these below.

Three new technological leaps

The appeal of linear models has always been their constant memory requirements and linear compute scaling.

However, as the Mamba 3 authors point out, there is “no free lunch”. By fixing the state size to ensure efficiency, these models are forced to compress all historical context into a single representation—the exact opposite of a Transformer’s ever-growing KV cache. Mamba-3 pulls three specific levers to make that fixed state do more work.

1. Exponential-Trapezoidal Discretization

State Space Models are fundamentally continuous-time systems that must be “discretized” to handle the discrete sequences of digital data.

Previous iterations relied on “Exponential-Euler” discretization—a heuristic that provided only a first-order approximation of the system.

Mamba-3 introduces a generalized trapezoidal rule, providing second-order accurate approximation. This isn’t just a mathematical refinement; it induces an “implicit convolution” within the core recurrence.

By combining this with explicit B and C bias terms, the researchers were able to remove the short causal convolution that has been a staple of recurrent architectures for years.

2. Complex-Valued SSMs and the “RoPE Trick”

One of the most persistent criticisms of linear models has been their inability to solve simple state-tracking tasks, such as determining the parity of a bit sequence.

This failure stems from restricting the transition matrix to real numbers, which prevents the model from representing “rotational” dynamics.Mamba-3 overcomes this by viewing the underlying SSM as complex-valued.

Using what the team calls the “RoPE trick,” they demonstrate that a complex-valued state update is mathematically equivalent to a data-dependent rotary embedding (RoPE) applied to the input and output projections.

This allows Mamba-3 to solve synthetic reasoning tasks that were impossible for Mamba-2.

3. MIMO: Boosting Arithmetic Intensity

The most significant leap in inference efficiency comes from the transition from Single-Input, Single-Output (SISO) to Multi-Input, Multi-Output (MIMO) SSMs.

In a standard SSM, the state update is an outer-product operation that is heavily memory-bound.By switching to a matrix-multiplication-based state update, Mamba-3 increases the “arithmetic intensity” of the model—the ratio of FLOPs to memory traffic.

This allows the model to perform more computation during the memory-bound decoding phase. Essentially, Mamba-3 utilizes the “idle” compute cores of the GPU to increase model power for “free,” maintaining the same decoding speed as its simpler predecessors.

What Mamba 3 means for enterprises and AI builders

For enterprises, Mamba-3 represents a strategic shift in the total cost of ownership (TCO) for AI deployments.

• Cost vs. Performance: By matched-parameter performance, Mamba-3 (MIMO) matches the perplexity of Mamba-2 while using half the state size. For enterprise deployment, this effectively doubles the inference throughput for the same hardware footprint.

• Agentic Workflows: As organizations move toward parallel, agentic workflows (like automated coding or real-time customer service agents), the demand for low-latency generation increases exponentially. Mamba-3 is designed specifically to prevent GPU hardware from sitting “cold” during these tasks.

• The Hybrid Advantage: The researchers predict that the future of enterprise AI lies in hybrid models. By interleaving Mamba-3 with self-attention, organizations can combine the efficient “memory” of SSMs with the precise “database” storage of Transformers.

Availability, licensing, and usage

Mamba-3 is not merely a theoretical research paper; it is a fully realized, open-source release available for immediate use with model code published on Github.

The project is released under the Apache-2.0 License. This is a permissive, business-friendly license that allows for free usage, modification, and commercial distribution without requiring the disclosure of proprietary source code.

This release is good for developers building long-context applications, real-time reasoning agents, or those seeking to reduce GPU costs in high-volume production environments.

Leading the State Space Models (SSM) revolution

The release was met with enthusiasm on social media, particularly regarding the “student-led” nature of the project. Gu, whose X/Twitter bio describes him as “leading the ssm revolution,” gave full credit to the student leads, including Aakash Lahoti and Kevin Y. Li

.Gu’s thread highlighted the team’s satisfaction with the design:

“We’re quite happy with the final model design! The three core methodological changes are inspired by (imo) some elegant math and methods.”

As agentic workflows push inference demand “through the roof,” the arrival of Mamba-3 suggests that the future of AI may not just be about having the biggest model, but about having the most efficient one.

Mamba-3 has successfully re-aligned the SSM with the realities of modern hardware, proving that even in the age of the Transformer, the principles of classical control theory still have a vital role to play.