AI Coding Live News

webAI CEO David Stout announced on X on November 4 (17:37 UTC) that the company is open-sourcing TwIL-8B. The official Hugging Face model card says it is post-trained from IBM Granite 4.2 8B with full-parameter SFT, GRPO/DAPO reinforcement learning and a length-aware RL stage, and ships in balanced and verbose variants. Self-reported scores include LiveCodeBench 81.6 (verbose), MATH-500 98.2 and IFBench 74.4, about 93% and 92% of the best frontier score, while GPQA Diamond is only 70.2. A 4.98 GiB Q4_K_M GGUF runs locally; the license is the webAI Non-Commercial License.

Zoom

Axiom released a Lean formalization of the entire Stacks Project on November 4 (official X post at 17:13 UTC, with over 2 million views). The roughly 7,500-page open-source algebraic geometry reference, started by Johan de Jong in 2005, becomes about 16 million lines of code, which Axiom calls the largest integrated, reusable formal math library ever built. Hierarchically coordinated AI agent swarms did the work and surfaced dozens of statements in the original that needed local fixes, with counterexamples. The library is open source and browsable side by side with the original text at stacks.axiommath.ai.

Zoom

On November 2, MIT CSAIL PhD student Rikiya Takehi posted "Weight-Space Programming" on X: a team (Takehi, Jayden Teoh, Alex Zhang, Omar Khattab) had GPT-6 Astra read and edit neural network weights directly, with no gradients or training. Their blog says Astra changed only 131,072 weights (0.016%) of Qwen3.5-0.8B, fixed its thinking loops and raised MATH-500 from 7.9% to 48.1%. The post drew about 24K likes and 1.77M views, and Ethan Mollick quoted it saying it is a big deal pending replication. All results are self-reported by the authors and not yet independently reproduced.

Zoom

a16z's Charts of the Week for Oct 30 (by Moses Sternstein) looks at the "Rack Spread": GPU rental prices are rising while token prices fall. The official a16z X account summed it up on Oct 31: since May, average token prices are down 67% and token spend is up more than 120%. By the article's napkin math, Ramp data gives roughly 122%, and Ornn's blended price multiplied by OpenRouter volume gives roughly 160%. A separate post says that since mid-July B300 rentals are up 106%, B200 up 66% and H100 up 17%. a16z also stresses that these indexes come from small, biased samples and do not show true total demand.

Zoom

Anthropic is extending its AI for Science program, launched in May 2025, from individual researchers to institutions: six science funders (including the US NSF, Simons, Sloan and Schmidt Sciences) will distribute Claude credits to researchers, six institutions (such as STScI, Sanger, the Crick and EMBL) will use credits in-house, and six public-good research projects (such as gnomAD, FlyWire and the Lean math library Tau Ceti) get direct support. Anthropic did not disclose the total value of the credits.

Zoom

AI math company Axiom announced on Oct 27 that swarms of dozens of AI agents formalized the Classification of Finite Simple Groups (CFSG; the original proof runs about 15,000 pages across 500+ articles by about 100 mathematicians) in the Lean proof assistant in 24 days, producing about 43.1M lines of Lean and 2.05M declarations, with the repo public on GitHub. The work surfaced 34 noteworthy deviations from the published record, and AxiomProver produced a new alternative argument for the Unbalanced Group Theorem.

Zoom

In an Oct 26 blog post, OpenAI said its internal models produced three new results: a proof of Sogge's local smoothing conjecture in every dimension (which implies the Kakeya and Stein restriction conjectures), a Lean-verified proof that the Exponential Time Hypothesis (ETH) is false, and a rigorous existence proof for four-dimensional QCD and related quantum field theories including pure Yang-Mills. The math and CS papers and Lean proofs are in the openai/math GitHub repo; the physics artifacts are being held back while an external group led by David Gross writes them up. All of this is OpenAI's own claim and still awaits independent verification.

Zoom

On Oct 26, Zerodha founder and CEO Nithin Kamath asked on X how people are using Kite APIs with AI agents now that they are free, and what they have built. He said AI coding agents such as Claude Code and Codex have made it much easier for traders and investors to build their own tools without much coding experience, and that Zerodha is seeing more Kite API usage because of these agents. Within about 10 hours the post drew more than 2,200 likes, over 2,000 bookmarks and 200+ replies. This is an X discussion and the CEO's own view, not a new product launch.

Zoom

GitHub Next officially released PDD Kit (Property-Driven Development Kit), establishing a formal verification toolkit tailored for autonomous coding agents. Addressing the pervasive reliability gap where agent-generated code passes isolated unit tests but crashes on subtle edge cases, PDD Kit combines property-driven specifications with automated formal solvers. Instead of testing arbitrary sample inputs, PDD Kit mathematically proves that agent code adheres to rigorous invariant specifications across every possible input, preventing regressions and elevating coding agents from probabilistic generation to provable correctness.

Zoom

Looped language models reuse physical Transformer blocks across logical depth to compress parameter footprints, but existing approaches entangle recurrence placement, input injection, and recurrent-state topologies. CMU and Fudan researchers published Part I of 'Towards Looped Models Done Right' (arXiv:2610.23036). Introducing a Prelude-Recurrent-Coda sandwich architecture, an 8.0B-resident / 794M-active Huginn MoE trained on 500B tokens matches a 32.0B-resident / 3.2B-active feedforward reference across MMLU, GSM8K, and MATH500.

Zoom

Tackling step-level verification in mathematical reasoning and in-context copying bottlenecks in Transformers, NYU/UCLA/Stanford researchers introduce LeanPRM (arXiv:2610.17697)—using the Lean 4 theorem prover to automatically label 152K steps across 10,634 natural-language proofs with zero human annotation, boosting Qwen3-4B step accuracy on LeanPRM-Bench from 48.2% to 74.7% (vs. 60.3% for PRM800K). Meanwhile, Cornell University's Copy-Paste Language Models (CPLMs, arXiv:2610.19247) add a lightweight Copy-Paste head (~3% compute overhead) that relieves the LM head gradient bottleneck, boosting LAMBADA accuracy by +2.5-4.0 points and allowing a w=128 sliding-window CPLM to match full-attention perplexity.

Zoom

Addressing rollout homogenization and zero-variance advantage collapse in group-relative reinforcement learning (GRPO), SERO (arXiv:2610.19397) restores learner-visited states to execute multi-step oracle continuations with policy-reachability weighting—boosting 7B ALFWorld OOD success by +11.20 points to 87.50%—while UNSW's EDEN (arXiv:2610.16105) appends a single anchored Gaussian-perturbed embedding position to a fraction of rollouts, lifting Qwen2.5-Math-7B AIME 2025 Pass@256 by +13.33 points (to 66.67%) without increasing rollout budgets.

Zoom

Topping today's Hugging Face Daily Papers are Waypoint-OPD (451 upvotes, arXiv:2610.21601) and FDE-Gym (230 upvotes, arXiv:2610.22350). Waypoint-OPD leverages an intermediate RL teacher checkpoint to gate endpoint policy-shift distillation on Qwen3-8B, 14B, and 30B-A3B students, achieving #1 accuracy across all six math and coding settings. Meanwhile, FDE-Gym benchmarks 13 frontier agent configurations across 95 realistic forward-deployed engineering delivery tasks, revealing that the best system resolves only 14 of 95 tasks when unstated customer constraints must be uncovered.

Zoom