Topic · AI

Open models

109 developments in the last 30 days.

Open models AI

69 developments this week +27 vs last week

Stage: Coverage

How momentum works

Opportunities

Bulk Pass: batch extraction over CSVs on the new fast, cheap modelsA CLI that tests a sample, picks the cheapest model that passes, then runs your whole file. Data-heavy small teams pay.First flagged 2026-10-17
Diffusion and Chinese flash models squeeze the small-model tier, and the DeepSeek ban splits buyers by originFast, cheap models for routine agent work. Inference platforms and regulated buyers pay, sorted by where the model came from.First flagged 2026-10-17
Amazon opens PyTorch on Trainium, and CUDA's moat is attacked at the framework layerRunning PyTorch code unchanged on non-Nvidia chips, paid for by hyperscalers and model labs buying cheaper compute.First flagged 2026-10-12
PRD Trace: a requirement-by-requirement check that an agent's PR did what the spec askedA GitHub app that maps each spec requirement to diff lines and tests and flags gaps. Small product teams pay per repo.First flagged 2026-10-20
Origin Check, now with a federal DeepSeek ban: model-origin scans for government vendorsFinds Chinese-developed models in code, gateways and laptops and produces attestations; fintechs selling to government pay.First flagged 2026-10-13
Port Desk: VB6 and Access ports for small businesses, now that OpenAI claims a legacy-language leapAgent-assisted rewrites of a small firm's VB6 or Access app as a web app. Owners pay a fixed price per port.First flagged 2026-10-11
Google open-sources self-tuning agent harnesses, and the app layer's hand-built moat thinsAgents that rewrite their own prompts, tools and memory for each task. Durable value moves to whoever owns the grader and replay data.First flagged 2026-10-19
Harness Tune: AGENTS.md edits proposed by Harness2, kept only if replays improveA CLI that reads failed coding-agent sessions, suggests instruction edits, replays them and keeps only the edits that win. Developers pay.First flagged 2026-10-18
Trail Report: a verifiable stolen-funds trace for small crypto theft victimsAn agent traces stolen funds to exchange deposit addresses and writes a hop-by-hop report for police and exchanges. Victims and small DAOs pay per report.First flagged 2026-10-19
China's export rule hands the open-weight enterprise market to US labsRegulated enterprises standardise on US open-weight models they host themselves. Revenue comes from fine-tuning, serving and support contracts.First flagged 2026-10-10
Labs bundle web search for agents, and standalone search APIs lose pricing powerModel labs now sell search and fetch alongside tokens. Agent builders pay less, and independent search-API vendors get squeezed.First flagged 2026-10-18
Search Shim: a drop-in search-and-fetch tool over DeepSeek's $0.75 API, with fallbackAn open-source MCP server and hosted router that defaults agent web search to DeepSeek's cheap API. Indie agent builders pay.First flagged 2026-10-18
Coding platforms swap default models, and lab revenue looks rentableData on how much frontier-lab revenue comes through app platforms that can change default models overnight. IPO investors would pay.First flagged 2026-10-11
Frontier MoE models stream experts from NVMe, adding pressure on HBMInference that pulls only the experts each token needs from DRAM and flash. Cost-sensitive operators running their own hardware pay.First flagged 2026-10-13
Rig Fit, now for expert streaming: size RAM and NVMe, not just VRAMA calculator and shared benchmark list that predicts tok/s for MoE models spread across VRAM, RAM and NVMe. Local-LLM buyers pay.First flagged 2026-10-11
Origin Swap: move a Qwen or DeepSeek fine-tune onto US open weights and prove parityA CLI that re-runs your fine-tune on Nemotron, Muse or Beam and compares results on your evals. Startups selling to US buyers pay.First flagged 2026-10-17
Agents that run inside the user's own browser get past bot walls, and sites start asking agents to identify themselvesTools that spot and set rules for agents acting inside a real user's logged-in browser. Banks and retailers pay.First flagged 2026-10-16
Escalate Only When Lost: small-model drafts with frontier checks for back-office agentsA local 24GB model drafts tool-call trajectories and a frontier model checks them only at risky points. Fintech ops teams pay.First flagged 2026-10-16
Like Audit: spot farmed likes on Hugging Face models before you trust a trending repoA browser extension and API that score whether a model's likes come from real accounts. ML engineers pay.First flagged 2026-10-18
Memory Write: a local memory server running a 3B-active writer modelAn open-source MCP memory server that uses Hy-Memory-30B-A3B for the write path. Developers pay for sync.First flagged 2026-10-17
Barge Test: interruption and disclosure testing for AI debt-collection callsSimulated borrowers who talk over, dispute and say 'stop' to AI collection voice agents. Collection agencies and voice vendors pay.First flagged 2026-10-17
Turn Kit: drop-in full-duplex turn-taking for open-source voice agentsA plugin that runs Tyger's answer/stop/chime-in model inside voice-agent frameworks. Indie voice-agent builders pay for the hosted version.First flagged 2026-10-17
Meta and Apple make the Mac a home for open frontier weightsMuse Spark Open on Apple silicon and Nvidia's 4GB–24GB Nemotron models let Mac apps skip API bills.First flagged 2026-10-16
Shelf Shots: consistent product photos from ten reference images on an 8GB GPUProduct-photo generator for Etsy and Shopify sellers that keeps the item exact across scenes. Sellers pay per batch.First flagged 2026-10-16
Zero-bill docs assistant: a 2B model running in the visitor's browserEmbeddable docs Q&A widget where inference runs on the reader's GPU via WebGPU. Maintainers pay nothing per question.First flagged 2026-10-16
Runway Bands: probabilistic MRR forecasts from Stripe data already in your PostgresSelf-serve forecasting that reads synced Stripe tables and shows indie SaaS founders revenue ranges. $19–49 a month.First flagged 2026-10-16
Nvidia ships its own 24GB models, and the local agent stack takes shapeSparse models that run agents on phones and 24GB cards. Device makers and privacy-bound enterprises pay.First flagged 2026-10-14
Statement Desk: local agent over your own bank exports, now that JPMorgan says noDesktop app plus MCP server that lets a local on-device model reason over your bank CSVs. Prosumers pay.First flagged 2026-10-14
On-device coached-scam check inside the payment flowA small model on the phone, inside the bank app, flags coached payments before they send. Neobank fraud teams pay.First flagged 2026-10-07
Scam Second Opinion: on-device check of suspicious chats and screenshotsBrowser extension and phone app that judge suspicious messages locally with small models. Worried families pay.First flagged 2026-10-15
Table Read: directed multi-voice audiobook drafts for indie authorsWeb app turning a manuscript into a directed, multi-character audio performance. Indie authors pay per finished hour.First flagged 2026-10-14
Self-spreading small models make consumer devices a containment marketNetwork-level detection and containment of AI models spreading across smart TVs, routers and IoT. ISPs and device makers pay.First flagged 2026-10-14
Bank examiners ask where the model came from, and model-origin audits formTools that inventory which base models sit inside a bank's own and vendor software, paid for by bank model-risk and third-party-risk teams.First flagged 2026-10-13
Model Origin Scan: find Chinese-developed models in a fintech's stackOpen-source scanner that lists every model, weight file and AI endpoint in a codebase, with lineage and origin. Fintechs pay.First flagged 2026-10-13
World-model labs become acquisition targets, and Decart sets the priceReal-time interactive world models for robotics, simulation and media, priced by big-tech acquirers.First flagged 2026-10-12
Weights Witness: proof your inference provider serves the model you validatedStatistically checks that a hosted open-weight endpoint is the exact model risk teams approved. Bank model-risk teams pay.First flagged 2026-10-10
Activation monitors move from research to a product you can buyProbes that read a model's internal activations to flag sabotage and deception, bought by anyone self-hosting agent models.First flagged 2026-10-09
Resettable enterprise replicas for training and evaluating agentsTeams improving agents need realistic, stateful copies of business systems that reset after every run, and that supply is becoming a category of its own.First flagged 2026-10-09
Clip Finder: local footage search for creators using EmbeddingGemma 2Desktop app that indexes a creator's raw footage locally and finds moments by description. Editors pay once.First flagged 2026-10-11
Decision Swap: codemod and router that moves classify-style LLM calls to decision modelsCLI that finds yes/no and pick-one LLM calls in your code and reroutes them to cheaper decision APIs; dev teams pay.First flagged 2026-10-10
Invoice Sense: does this invoice fit this business, checked on every paymentFlags invoices whose goods, prices or routes don't fit the payer's business. FCA-registered EMIs and payment fintechs pay.First flagged 2026-10-10
Haiku-priced E2E testing for indie web appsPlain-English end-to-end tests run by a cheap computer-use model against your staging site on every deploy, which removes most of the work of writing Playwright tests.First flagged 2026-10-09

Developments

Last 30 days