The Hub for AI Coding Models, Tools & Workflows

On [2026-09-30](https://technode.com/2026/09/30/ant-group-launches-ling-3-1-flash-with-560-billion-parameters/) InclusionAI (Ant Group) announced **Ling-3.1-flash**: ~**560B** total / ~**25B** active MoE hybrid-reasoning model, designed for up to **1M** context (trial/gateway routes often expose **256K–262K**). Weights are **not** on Hugging Face yet. [Vercel AI Gateway](https://vercel.com/changelog/ling-3-1-flash-is-now-available-on-ai-gateway) and [OpenCode Zen](https://opencode.ai/docs/zen/) list limited-time free access (~through **2026-10-13**). Vendor-reported scores include SWE-Pro **65.39**, Terminal-Bench 4.0 **40.40**, CyberGym **87.90**—not independently reproduced.

On [2026-10-02](https://www.anthropic.com/news/claude-frontier-academy) Anthropic launched **Claude Frontier Academy** with a **$100M** commitment to train **10,000** enterprise **Frontier Deployed Engineers (FDEs)** by end of **2027**. First cohorts include Accenture, Bain, Capgemini, CBA, Deloitte, McKinsey, Morgan Stanley, and Novo Nordisk. Path: multi-day in-person simulated deployment → **Claude Resident Engineer** badge → **12-week** residency on a named Claude use case → **Claude Frontier Deployed Engineer** badge (first expected early 2027). Nomination-only—not a public self-serve course.

On [2026-10-01](https://allenai.org/blog/olmocore3) Ai2 released **Olmo-core 3**, an open MoE training stack ([GitHub](https://github.com/allenai/OLMo-core), PyPI `ai2-olmo-core`). It moves from FSDP to **DDP + expert parallelism**; on 8×B300 a 47B MoE hits ~**2.7×** throughput (52k vs 19.4k tok/s/GPU). Systems benches include **1.2T** on 512 GPUs at up to **858** useful-model TFLOP/s/GPU and a DeepEP v2 capacity probe at **2.38T**. Report: [olmocore3](https://allenai.org/papers/olmocore3).

On [2026-09-30](https://cohere.com/blog/embed-5) Cohere launched **Embed 5** with **Pro** and **Fast** tiers that share one embedding space (index with Pro, query with Fast). Multimodal text/image/fused inputs, **128K** context, 100+ languages; API ids `embed-v5.0-pro` / `embed-v5.0-fast`. Blog: ViDoRe V3 Pro avg **85.8** (+8.8 vs Embed 4); text pricing **$0.12 / $0.08** per M tokens. Available on Cohere API, Model Vault, Azure Foundry, SageMaker; vLLM for private deploy.

On [2026-10-01](https://blog.cloudflare.com/ai-search-ga/) Cloudflare made **AI Search generally available**: managed index/retrieval (Workers AI + Vectorize + R2 + Browser Run) ships native multimodal embeddings ([Qwen3-VL-Embedding](https://developers.cloudflare.com/ai-search/configuration/models/supported-models/)), OCR for scanned PDFs, 10 MiB text/OCR PDF limits, hybrid search by default, and [billing from 2026-11-01](https://developers.cloudflare.com/ai-search/platform/limits-pricing/) with a free allotment.

On [2026-10-02](https://huggingface.co/blog/allenai/astabrief) Ai2 open-sourced **AstaBrief 8B** (Qwen3-8B base): one-pass cited scientific reports from a question + retrieved snippets. Powers Asta **Fast** mode vs Claude Thinking. End-to-end report time averages **51.1s** vs **178.5s** (~3.5×). Weights: [allenai/AstaBrief_8B](https://huggingface.co/allenai/AstaBrief_8B) (updated 2026-10-02).

On [2026-10-02](https://blog.cloudflare.com/introducing-web-search-api/) Cloudflare launched **Web Search API** (open beta): agents query the live web through [AI Gateway](https://developers.cloudflare.com/ai-gateway/) and get structured titles/URLs/snippets. Default [Ceramic.ai](https://developers.cloudflare.com/web-search/providers/) ($0.25/1k, ZDR); Exa ($7) and Linkup ($5) also available. Partners meet Verified-bot rules; REST, Workers `env.AI.websearch()`, BYOK, list pricing with no markup.

On [2026-10-01](https://strandsagents.com/blog/introducing-strands-decider/) AWS Strands Labs released **Strands Decider 2B** ([`strands-labs/strands-decider`](https://github.com/strands-labs/strands-decider), Apache-2.0; weights [HF hobson-v19](https://huggingface.co/StrandsAgents/strands-decider-2B-hobson-v19)): LM head replaced by an option-scoring head. JevBench public **167/231 ≈72.3%**, Brier ~0.35; ~**115ms** median on RTX 3090. `pip install strands-decider` for routing, tool choice, and before_tool_call gates.

On [2026-10-01](https://blog.cloudflare.com/cloudflare-os/) Cloudflare open-sourced **Cloudflare OS** ([github.com/cloudflare/cloudflare-os](https://github.com/cloudflare/cloudflare-os), Apache-2.0): a browser agent workspace with company context/skills, modifiable Gadgets, and a **Gatekeeper** capability-security layer (zero default access, observation-following policy, deferred simulated approvals). Deploy via [starter](https://github.com/cloudflare/cloudflare-os-starter) or https://os.cloudflare.app/deploy; inference through AI Gateway.

On 2026-10-01 Google’s [Project Suncatcher](https://blog.google/innovation-and-ai/models-and-research/google-research/project-suncatcher-prototype/) prototype, built with Planet, reached orbit on SpaceX Transporter-18 (Falcon 9). Contact confirmed. Focus: TPU survival plus **radiative/heat-pipe cooling in vacuum** (no airflow), plus radiation/launch stress data. Background: [Facts](https://blog.google/innovation-and-ai/models-and-research/google-research/google-project-suncatcher-facts/).

On [2026-10-01](https://aws.amazon.com/blogs/aws/announcing-aws-well-architected-agent-an-ai-powered-intelligence-to-optimize-your-cloud-environment-preview/) AWS announced public preview of the **Well-Architected Agent**: it correlates metrics, configs, and topology against Well-Architected practices across 65+ services, prioritizes by declared business goals, and ships console / updated IaC (Terraform, CDK, CloudFormation) / CLI remediation. Available in us-east-1, us-east-2, us-west-2; delivered via AWS Support.

On [2026-10-01](https://developer.nvidia.com/blog/build-applications-on-nvidia-bluefield-faster-with-nvidia-doca-agent-skills/) NVIDIA published **DOCA AI agent skills** on [NVIDIA/skills](https://github.com/NVIDIA/skills): SKILL.md packs with real API signatures, hardware capability checks, and build constraints for Flow, GPUNetIO, PCC, RDMA, and more. Vendor 65-prompt eval: 19% checklist items without skills vs 100% with; RDMA demo used 73% less handwritten code and 46% fewer hardware commands. Installable into Claude Code, Codex, and other coding agents.

On [2026-10-01](https://www.digitalocean.com/blog/introducing-agent-droplets) DigitalOcean launched **Agent Droplets** for Managed Agents (public preview): Pro $50/mo (15% off) and Team $200/mo (20% off) bundling Harness Runtime microVMs, DO-hosted open-model inference (Kimi K3, GLM 5.3), session storage, and Action Gateway (16,000+ tools). Works with Claude Code, Codex CLI, OpenCode. Paired with prepaid Inference and Agents Balance.

GitHub’s [2026-10-01 Changelog](https://github.blog/changelog/2026-10-01-dynamic-workflows-in-copilot-cli-and-the-copilot-app/) launched **Dynamic Workflows** in public preview for Copilot CLI, the Copilot app, and the SDK: **code-defined** reusable orchestration (sequential/parallel steps, pause checkpoints, subagent cross-checks)—unlike Autopilot or `/fleet`, where Copilot invents the plan each run. CLI needs `--experimental` or `/experimental on`; available on all Copilot plans.

Cursor’s official [Models & Pricing](https://cursor.com/docs/models) and dedicated docs now list Z.ai **GLM 5.3** and **GLM 5.3 Flash** as addable third-party models: 1M context, Low/High/Max reasoning (High default), Other Models pool billing. Table rates: GLM 5.3 **$1.4/$0.26/$4.4**, Flash **$0.15/$0.029/$0.5** per MTok—same as 5.2, no long-context surcharge. Enable in Settings → Models (not the BYOK custom-endpoint path).

Over ~7 days, Claude Sonnet 5.5, GPT-6.1 Sol, Gemini 4 Argon, JetBrains Air in IDEs EAP, and GitHub Copilot Computer Use landed in the same window. Search and HN discourse shifted from single-release chase to pairwise selection on price bands, terminal agents, IDE harnesses, and computer-use. Meta industry brief for readers without X.

[@Cloudflare](https://x.com/Cloudflare/status/2105747536510099540) (2026-10-01, #BirthdayWeek) launched its first in-house decision models **Clef** (27B) and **Clef-flash** (9B): hosted on Workers AI, [Apache 2.0](https://huggingface.co/Cloudflare/clef) weights, Jev System One–compatible; official blog cites ~**2.5× / 13×** median latency vs Jev and **2.2s vs 4.7s** in a threat-intel Browser Run workflow.

Gemini post-training/RL engineer Qiao Zhang ([@zhangqiaorjc](https://x.com/zhangqiaorjc/status/2105509406058463657), 2026-10-01), quoting the [Gemini 4 Argon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/) launch, said mini-breakthroughs are accelerating: an overnight elimination of trainer–sampler mismatch and another ~10% MFU gain—RSI in full swing. Recipes are not in the official blog; access remains via [Fairwind](https://deepmind.google/fairwind-program/).

Official [@bot](https://x.com/bot/status/2105713240701538538) (2026-10-01): Grok Bot can now suggest ways to help without you needing to ask. Fits the [Chief of Staff](https://cursor.com/docs/grok-bot/use-cases) pattern plus cloud computer, skills, and routines; consequential actions stay behind Approvals/Auto Review. Get the app at [x.ai/bot](https://x.ai/bot).

On 2026-10-01 GitHub announced Computer Use in public preview for Copilot CLI and the Copilot app on macOS and Windows: agents can drive local desktop apps via accessibility tree/screenshots, clicks, typing, scroll/drag, and cross-app workflows—targeting GUI-only legacy software. Off by default; enable with `/computer on` or Settings; enterprise policy can block it.