AI news5
TypeSafe raises $870 million at a $7.5 billion valuation
TypeSafe confirms a $870 million Series A led by Andreessen Horowitz, valuing the company at $7.5 billion. The company says around a third of Fortune 500 businesses use Jev, its model for structured decisions with confidence estimates. It plans more machine-native models and enterprise features.
typesafe.ai ↗Why it matters
This much larger round follows September’s $40 million announcement, funding an expansion of decision models that software can act on directly.
Microsoft launches a decision model at $0.042 per million input tokens
Microsoft-Decision-1 is available through Microsoft Foundry and OpenRouter, returning calibrated scores for fixed answer choices. Input costs $0.042 per million tokens and output is free. Microsoft reports that Xbox Research classified more than 10,000 feedback entries at over 14 times the speed and 200 times lower cost than GPT-6 Sol.
commandline.microsoft.com ↗Why it matters
The new decision-scoring release is a separate milestone from Microsoft’s earlier agent playbook, with an explicit price for routing and quality-control workloads.
Cloudflare adds audio and video to Clef-omni
Cloudflare’s new open-weight Clef-omni accepts audio, video, images and text in one decision-scoring call. It is built on Qwen3-Omni-30B-A3B, with speech-generation components removed. Cloudflare reports scoring a 21-second video with sound in about 1.5 seconds.
blog.cloudflare.com ↗Why it matters
This adds a multimodal release to the earlier Clef family, allowing developers to evaluate media without first chaining separate transcription and image models.
KernelAI brings 53 local models to iPhone and Android
KernelAI’s Product Hunt launch on 10 October introduces its offline assistant for iPhone and Android. Its current catalogue lists 53 models from 15 labs, with downloads from 112 MB to 3.11 GB. The developer says document chat, image questions and dictation run on the device; web search is optional and GGUF imports are experimental.
kernelai.app ↗Why it matters
The launch offers a concrete way to test private mobile AI, with model availability constrained by each phone’s memory budget.
Nextech adds two AI platforms for 24-hour patient access
Nextech announced multi-year partnerships with Assort Health and EliseAI for specialty healthcare practices on 9 October. The integrations connect AI patient access with scheduling, intake and follow-up workflows, including outside office hours. The announcement describes joint engineering and shared roadmaps; it does not report measured results from these new partnerships.
nextech.com ↗Why it matters
Connecting patient-access agents to existing practice systems makes workflow integration a central part of the product, rather than relying on a standalone chatbot.
AI in the nonprofit sector4
o9 opens a $250,000 digital-inclusion grant for nonprofits worldwide
AI software company o9 will award $250,000 to one nonprofit through its Digital Tomorrow Grant 2026. Applications from nonprofits worldwide close on 31 October, including joint applications. Eligible projects cover digital access, literacy and technology-enabled services; an AI project is not required.
o9solutions.com ↗Why it matters
This provides funding for the digital foundations that underserved communities need before they can benefit from AI tools.
Bonfils-Stanton opens grants up to $60,000 and clarifies AI-writing policy
Bonfils-Stanton’s quarterly round opened on 9 October and closes on 10 November, with decisions in January 2027. Operating and project grants are capped at $60,000 for eligible Denver arts nonprofits and fiscally sponsored groups. Applicants may use AI to support grant writing, but the foundation says it does not use AI to review applications.
bonfils-stantonfoundation.org ↗Why it matters
A live funding round gives applicants a clear AI policy while preserving the need for their own programme knowledge and authentic voice.
BetaNYC publishes a searchable transcript of a 10-hour AI hearing
Civic technology organisation BetaNYC has published a free searchable transcript of the New York City Council’s 5 October AI hearing. It includes speaker names, timestamps, data files and code for the ten-hour session covering AI risks and ten related bills. BetaNYC warns that the transcript is machine-generated and quotations should be checked against the video.
beta.nyc ↗Why it matters
Civil society can inspect a substantial public hearing through reusable evidence, without treating proposed legislation as enacted law.
AIoD prepares funding of up to €60,000 for public-sector AI providers
AI-on-Demand’s second call is scheduled to open in the second half of October, with a 30 November deadline at 5 PM CET. Eligible providers include research organisations and academia in EU or DIGITAL Europe associated countries, with mature AI products at TRL 7 or above. The two-stage programme offers providers up to €60,000 and public-sector adopters up to €15,000; applications are not yet open.
aiodp.ai ↗Why it matters
Research institutions have time to check eligibility and arrange a public-sector deployment partner before the submission window opens.