This frozen page shows Augur's claims and source links for one sent dispatch. Stored spot-checks appear only where the frozen edition supports them; absence is not presented as verification.
Earlier numbers point the same direction: the company has said Claude wrote more than 80 percent of its merged code as of May, and lines merged in 2026 run about 8x the 2021 to 2024 average Anthropic Engineering Nate Jones Import AI.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic observed an 8x increase in lines of code merged in 2026 compared to the average of years 2021-2024.
Claim 1350Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 2
Changes are getting larger as well as more frequent, and a 50-million-line Ruby migration that Stripe and Anthropic say took one day instead of an estimated two-plus months is exactly the kind of output someone still has to check Simon Willison's Weblog AI Engineer.
Assertion status: No spot-check verdict is published for this assertion.
Coding agents are constructing increasingly large and sophisticated changes.
Claim 7455Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Stripe and Anthropic reported migrating a 50-million-line Ruby codebase in a single day using AI, a task they had estimated would take a team over two months.
Claim 72340Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
The Muse Wonder and Muse Wonder Full models are the first frontier model releases from Meta since Alexandr Wang restructured the company's AI efforts in August of the previous year.
Claim 76939Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 4
TechCrunch's report has Wonder Full reserved for Muse Code subscribers and absent from the API (the interface other software uses to call a model), while Latent Space lists both models as reachable through the Meta Model API and Muse Code Latent Space.
Assertion status: No spot-check verdict is published for this assertion.
Meta launched the Muse Wonder model family, consisting of Muse Wonder and Muse Wonder Full, described by the source as codenamed "Watermelon."
Claim 77305Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 5
On September 30 Google limited Gemini 4 Argon, its newest top model, to government users and trusted cyber defenders in its Fairwind Program, and said it would widen access to developers, enterprises, and consumers as soon as it had feedback from those early testers Google DeepMind Blog Latent Space.
Assertion status: No spot-check verdict is published for this assertion.
Google DeepMind announced the release of its new frontier model, Gemini 4 Argon.
Claim 70573Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Google DeepMind intends to make Gemini 4 Argon available to developers, enterprises, and consumers as soon as possible after gathering feedback from early testers.
Claim 70576Label: forecastProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 6
Argon is now open to every developer through the API in AI Studio, and to Pro and Ultra subscribers in the Gemini app and AI Mode in Search Google DeepMind Blog.
Assertion status: No spot-check verdict is published for this assertion.
Gemini 4 Argon is globally available for developers and enterprises via the API in AI Studio and the Gemini platform.
Claim 77451Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 7
The one outside datapoint, Artificial Analysis's 50 percent score on its AA-Omniscience test (an independent benchmark firm's knowledge test), is a single score with no comparison in this evidence, so it cannot tell you how Argon ranks on your work; it is a reason to test rather than assume Latent Space.
Assertion status: No spot-check verdict is published for this assertion.
Google DeepMind introduced the Gemini 4 Argon model for coding, enterprise knowledge work, and cyber defense.
Claim 71798Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 8
By Apple's accounting, the added language is worth as much as two to three times the scarce target text when measured by validation loss (a standard error score on held-out text) and two to thirteen times when measured by downstream accuracy (how well the model does on real tasks afterward) Apple Machine Learning Research.
Assertion status: No spot-check verdict is published for this assertion.
In data-constrained bilingual pre-training settings, mixing in data from a high-resource auxiliary language yields larger improvements in validation loss and downstream task accuracy compared to aggressive hyperparameter tuning.
Claim 77260Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Auxiliary data mixing boosts performance by an amount equivalent to 2–3 times the unique target data on validation loss and 2–13 times on downstream task accuracy.
Claim 77262Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 9
This fits Apple's earlier finding that scarce target text can be reused 15 to 20 times inside a mixture, and Cohere's data-mixing recipe for in-language reasoning Apple Machine Learning Research Cohere Blog.
Assertion status: No spot-check verdict is published for this assertion.
Scarce target corpora can be reused 15–20 times in mixture training, with the optimal number of repetitions depending on target data size, compute budget, and model scale.
Claim 49528Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Too little target data in a mixture underexposes the model to the target domain, while too much target data yields diminishing returns and eventual overfitting due to excessive repetition.
Claim 49531Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Tiny Aya L2-Thinker was trained using a data mixing strategy of English reasoning data, multilingual reasoning data, and multilingual non-reasoning data.
Claim 73616Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
The Muse Wonder and Muse Wonder Full models are the first frontier model releases from Meta since Alexandr Wang restructured the company's AI efforts in August of the previous year.
Claim 76939Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 11
- Independent evaluations of Argon beyond Artificial Analysis's 50 percent AA-Omniscience score are the numbers to set against Google's own claims. Latent Space
Assertion status: No spot-check verdict is published for this assertion.
Google DeepMind introduced the Gemini 4 Argon model for coding, enterprise knowledge work, and cyber defense.
Claim 71798Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 12
- Scion is in-house today; whether it ever becomes available outside Anthropic is an open question, and if it did, large engineering teams would have a new tooling option to weigh. Anthropic Engineering
Assertion status: No spot-check verdict is published for this assertion.
Anthropic averaged approximately 2,500 code changes merged per day in May 2026.
Claim 77235Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 13
- Ollama's 0.40.4 release lets you run decision models (a newer kind of model built to pick between set answers) straight from the terminal with one command, and separately adds Kev decision models to its Apple-silicon runner, which makes a local-versus-cloud comparison cheap to run. Ollama Releases
Assertion status: No spot-check verdict is published for this assertion.
The Ollama v0.40.4 release, published on 2026-10-14, enables the use of decision models directly from the terminal via the `ollama run` command.
Claim 77769Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 14
- Nate Jones points to an odd split in OpenAI's unreleased model: it could reconstruct a roughly 16,000-line program just by watching how it behaved, yet it botched a request to adapt an existing library. Test agents on unfamiliar customization work before trusting them with it. Nate Jones
Assertion status: No spot-check verdict is published for this assertion.
OpenAI released more than 700 mathematical research manuscripts generated by an internal, unreleased model, including results supported by machine-checked proofs.
Claim 77370Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
The author forecasts that AI capabilities may become more uneven, with exceptional abilities improving much faster than weaker ones, due to training feedback concentrating gains where it is cheap and plentiful.
Claim 77373Label: forecastProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 15
Anthropic's merge counts are self-reported, and the material available here gives no definition of a change's size and no defect figure; the 80 percent authorship number reaches this brief through a third party Nate Jones.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic reports that Claude has authored more than 80% of the code being merged into its code base as of May.
Claim 72446Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded