Today in DeAI: Chutes ends subscription renewals for pay-as-you-go billing, Palo Alto Networks agrees to buy open-weight platform Applied Compute for $4.5 billion in stock, and Tencent opens the Hy4 flash weights with a restrictive license.
Chutes ends subscriptions, goes pay-as-you-go only
The largest inference subnet on Bittensor has stopped selling monthly plans, and its two newest beta models — Qwen3.8 Max and DeepSeek V4.1 Pro — are PayGo-only, per the operator's October 26 announcement and its now-updated pricing page. The tiers being retired were $3, $10, and $20 a month; the pricing page now states "Monthly plans are no longer sold on chutes.ai." Why it matters: the lowest-commitment entry point to a major decentralized network is gone, and the new flagship tier lacks TEE support for now. (Chutes announcement) Our full coverage
Palo Alto Networks buys Applied Compute for $4.5B in stock
The security vendor announced a definitive all-stock agreement valued at $4.5 billion, subject to adjustments, to acquire the year-old startup that customizes and securely deploys open-weight models; Reuters and WSJ confirm the terms. Applied Compute's CEO Yash Patil becomes Palo Alto's Co-Chief AI Officer, and the deal is expected to close in PANW's fiscal Q3 2027. Why it matters: a $30B+ security company paying $4.5B for open-weight deployment tooling is the strongest enterprise-demand datapoint for secure open models this month — deal terms are company-announced and wire-confirmed, while the strategic framing is company narrative. (Palo Alto Networks press release, Reuters)
Tencent opens Hy4 flash weights — license is not Apache 2.0
The weights for Tencent's 220B-parameter, 12B-active multimodal MoE are live on Hugging Face with a 1M-token context, and community GGUF quants followed within hours. DeAI read the LICENSE file: the Tencent Hy License v1.0 adds MaaS commercial terms above $50M revenue or 100M monthly active users, a change from the Hy4 preview's Apache 2.0. Why it matters: the self-hosting math of a 12B-active model at 1M context is genuinely interesting, and the license terms decide who can actually serve it. (Hugging Face model card, LICENSE file) Our PULSE coverage
Mistral ships Vibe Desktop with local and remote agent execution
Mistral AI's coding agent is now a native macOS and Windows app that runs agents locally on the device or remotely, at the same time, with Linux listed as coming soon on the product page. Why it matters: a hosted lab embracing on-device execution normalizes the split workflow — sensitive work local, heavy bursts remote — that private-inference stacks are built around. (Mistral announcement, product page) Our PULSE coverage
Google ships Gemini 4 Neon at $3/$15 per million tokens
Google's fourth new Gemini model in a month is live through the API at $3 per million input tokens and $15 per million output tokens, priced above its predecessors' introductory rates, with a ~1M-token context window, per The Decoder's report of Google's announcement. Why it matters: release velocity itself is the story for builders pricing against closed APIs — the newest flagship costs more than the one it follows, so no single launch is a stable anchor. (The Decoder)
Watching tomorrow
The 35% Chutes discount on GLM-5.3 and DeepSeek V4.1 Flash expires November 23 — expect the pricing table to be the only rate that matters after that, and watch for Qwen3.8 Max and DeepSeek V4.1 Pro per-token rates to appear as the models exit beta.
Sources
- Chutes pricing page — model rates, Private Chutes, and billing FAQ — Chutes
- Chutes Rolls Out Qwen3.8 Max and DeepSeek V4.1 Pro as Subscriptions End — Tao.media
- Palo Alto Networks to Acquire Applied Compute, Bringing Secure Open-Weight AI to the Enterprise — Palo Alto Networks
- Palo Alto raises annual forecast, signs $4.5 billion deal for Applied Compute — Reuters
- tencent/Hy4-flash model card — Hugging Face
- Mistral Vibe product page — Mistral AI
- Google ships Gemini 4 Neon, its fourth new Gemini model in a month — The Decoder
About DeAI
DeAI is an independent publication covering open-weight AI models, private inference, and decentralized infrastructure — the tools for running AI you actually control. We test providers on price, privacy, and refusal behavior and publish the numbers, not the vibes. DeAI is powered by Morpheus (mor.org), a decentralized inference marketplace, and covers it on the same terms as every other provider.
Powered by Morpheus and StrandCMS
Morpheus is a decentralized inference marketplace, covered on the same terms as every other provider — we rank it wherever the data lands. StrandCMS is the open-source, agent-first framework this site is built on.
