{"access":"public \u2014 no key","articles":[{"canonical":true,"description":"Ten days in July 2026 produced five flagship AI releases \u2014 the largest open model ever (Kimi K3), the leading U.S. open model (Inkling), and three closed frontiers (GPT-5.6 Sol, Claude Fable 5, Grok 4.5). We scored them all on the same alphabet test. Open or closed, every one failed \u2014 and the open-vs-closed gap turns out to be a rounding error next to the real divide.","product":"l26","slug":"open-vs-closed-alphabet-test-july-2026","title":"Kimi K3, Inkling, GPT-5.6 Sol, Claude Fable 5, Grok 4.5: Five July Flagships, One Alphabet Test \u2014 Open and Closed AI Both Failed","url":"https://l26.ai/open-vs-closed-alphabet-test-july-2026"},{"canonical":true,"description":"In July 2026 we gave the three newest frontier AI models the simplest test in language AI: recite a language's complete operating alphabet.","product":"l26","slug":"gpt56-fable5-grok45-alphabet-test","title":"GPT-5.6, Fable 5, Grok 4.5 Compared on Alphabet Tests \u2014 and the Results Are In: All Three Failed","url":"https://l26.ai/gpt56-fable5-grok45-alphabet-test"},{"canonical":true,"description":"GPT-5.6 Sol and Grok 4.5 shipped in July 2026. Within days we scored them on L26, the operating-alphabet benchmark. The newest generation did not close the foundation gap \u2014 one flagship scored below its predecessor \u2014 and the data says the gap will not close on its own.","product":"l26","slug":"frontier-generation-alphabet-test","title":"The newest frontier models just took the alphabet test. The gap didn't close.","url":"https://l26.ai/frontier-generation-alphabet-test"},{"canonical":true,"description":"Gemini 3.6 Flash leads the L26 board. Google then shipped 3.7 Flash and 3.8 Flash \u2014 and both scored LOWER than the model they replaced. On Bemba the newest Flash recovers a third of what its own predecessor recovered, while still scoring a perfect 26/26 on English. Deliberation time halved across the same releases.","product":"l26","slug":"gemini-flash-generations-alphabet-test","title":"Google Shipped Two Newer Flash Models. Both Got Worse at the Alphabet Test.","url":"https://l26.ai/gemini-flash-generations-alphabet-test"},{"canonical":true,"description":"Nine models from five labs \u2014 OpenAI, DeepSeek, Z.ai, Alibaba, Meta and NVIDIA \u2014 arrived in the first days of September 2026. None displaced a single model from the top five, a July release still leads, and at two labs independently the newer model scores worse than the one it replaced. Every one of them aces English.","product":"l26","slug":"september-cohort-alphabet-test","title":"Nine New Frontier Models Took the Alphabet Test. The Best One Came Ninth.","url":"https://l26.ai/september-cohort-alphabet-test"},{"canonical":true,"description":"OpenAI's GPT-6 Astra, released September 3, 2026 and widely described as possible AGI, scores a perfect 26/26 on the English alphabet. Asked for the syllables of Luvale, it declined five times out of five. Handed the list, it rebuilt Bemba, Kinyarwanda and Luvale perfectly. A language-equity result on the most basic test there is.","product":"l26","slug":"gpt6-astra-alphabet-test","title":"They Are Calling It AGI. It Could Not Recite an Alphabet.","url":"https://l26.ai/gpt6-astra-alphabet-test"},{"description":"When an AI lab encounters a Full Syllable Inventory for the first time, the question they ask is what is the licensing fee. That is the wrong frame. FSIs are not a dataset you train on. They are the substrate every downstream operation references \u2014 closed, enumerable, standardised. Subscribe accordingly.","product":"fsi","slug":"fsis-are-not-data-theyre-infrastructure","title":"Full Syllable Inventories are not data, they're infrastructure!","url":"https://fsi.bantunomics.com/proofset/fsis-are-not-data-theyre-infrastructure.html"},{"description":"The syllable is the unit. Not the letter. By age 6 a Bantu child has internalised the syllabary of their language \u2014 the same inventory a frontier LLM in 2026 cannot enumerate. One layer off, and the error propagates through every prediction.","product":"fsi","slug":"how-a-bantu-child-learns-to-read","title":"How a Bantu child learns to read","url":"https://fsi.bantunomics.com/proofset/how-a-bantu-child-learns-to-read.html"},{"description":"There is a well-known African proverb: it takes a village to raise a child. AGI is no different. If it is going to serve all of humanity, it literally takes a global village to raise it.","product":"fsi","slug":"it-takes-a-village-to-raise-agi","title":"It Takes a Village to Raise AGI","url":"https://fsi.bantunomics.com/proofset/it-takes-a-village-to-raise-agi.html"},{"description":"L26 asks a basic question: can a model list every building block a language is made from \u2014 its complete alphabet? Frontier models nail English but fail the Bantu alphabets they","product":"fsi","slug":"operating-alphabet-benchmark","title":"The alphabet test every frontier model fails \u2014 L26 Operating-Alphabet Benchmark | BantuNomics","url":"https://fsi.bantunomics.com/proofset/operating-alphabet-benchmark.html"},{"description":"Imagine the AGI did not know the 26 letters. It would still write fluently \u2014 and still fail to sound out a new word, spell, or teach a child to read. You would stop calling it AGI on the spot. That is the model's literal condition today for 500+ Bantu languages: it does not have the alphabet, and it cannot feel that it's missing.","product":"fsi","slug":"the-alphabet-test-for-agi","title":"The alphabet test for AGI","url":"https://fsi.bantunomics.com/proofset/the-alphabet-test-for-agi.html"},{"description":"Every Bantu syllable fits one equation \u2014 \u03c3 \u2192 N\u2080C\u2080H\u2080G\u2080V. We have now tested it across 500+ Bantu languages, and it still holds, with no structural change. That law is why a Full Syllable Inventory takes the shape it does, why validating the parts settles the whole, and why a closed native inventory can tell you which words are borrowed.","product":"fsi","slug":"the-bantu-syllable-equation","title":"The Bantu Syllable Equation","url":"https://fsi.bantunomics.com/proofset/the-bantu-syllable-equation.html"},{"description":"Bantu languages have been spoken for thousands of years. The writing system is barely a hundred years old \u2014 and it leaves out the part that decides meaning. We named this gap so the AI labs could see it.","product":"fsi","slug":"the-flat-text-problem","title":"The Flat Text Problem","url":"https://fsi.bantunomics.com/proofset/the-flat-text-problem.html"},{"description":"AI labs have a name for it: jagged intelligence. The polite term for the blindness is unknown unknowns. The Bantu family used to be one of those. It isn't anymore \u2014 we have named it, mapped it, and measured the gap. And we are starting with the most foundational manifestation: the syllable layer every higher-layer Bantu AI failure mode stacks on top of.","product":"fsi","slug":"the-known-unknown","title":"The known unknown","url":"https://fsi.bantunomics.com/proofset/the-known-unknown.html"},{"description":"Run the same alphabet test twice \u2014 once closed-book, once with tools \u2014 and the gap between the two scores tells you whether a model knows a language or can merely look it up. A frontier model recovered toned Mandarin Pinyin almost completely with tools. On three Bantu languages it barely moved.","product":"fsi","slug":"the-tool-lift-test","title":"We gave a frontier model the internet. It rescued Mandarin and not Bemba. | BantuNomics","url":"https://fsi.bantunomics.com/proofset/the-tool-lift-test.html"},{"description":"Tone in a Bantu language is not decoration \u2014 it is meaning. And tone does not attach to the letter or the morpheme. It attaches to the syllable. If a frontier model cannot enumerate the syllables of a Bantu language, it cannot understand its tone. The two failures are the same failure.","product":"fsi","slug":"tone-lives-in-the-syllable","title":"Tone lives in the syllable","url":"https://fsi.bantunomics.com/proofset/tone-lives-in-the-syllable.html"},{"description":"You've seen the model can't detect its own missing foundation. Here's what that hole costs you, in numbers you already track: tokens you shouldn't need, products that underperform in the markets you paid most to enter, and an entire error class your text-based evals cannot see. Against any one of them, the subscription is small.","product":"fsi","slug":"what-the-missing-fsi-costs-you","title":"What the missing FSI costs you","url":"https://fsi.bantunomics.com/proofset/what-the-missing-fsi-costs-you.html"}],"count":17,"ecosystem":"BantuNomics","note":"Free BantuNomics thought-leadership across the ecosystem \u2014 Bantu language infrastructure, FSIs, the operating-alphabet benchmark, tone, the flat-text/AGI argument. Each article's `product` names its home subdomain.","sources":["https://l26.ai/api/articles","https://fsi.bantunomics.com/api/articles"]}