Independent/Reader-funded/Infrastructure, not tokens
DeAINEWS

AI you control — open models, private inference, and the networks that run them.

Daily Brief

DeAI Daily Brief — 16 November 2026

Today in DeAI: Meituan ships LongCat-2.5 open weights under MIT, OpenAI's usage-policy rewrite lands Nov 20, and GPT-5.6 Sol promo pricing expires Nov 21.

DeAI is powered by Morpheus (mor.org). We cover competing providers on the same terms — see our methodology.

A datacenter rack aisle with accelerator cabinets and an amber status LED, illustrating Meituan's LongCat-2.5 open-weights model reportedly trained on Huawei Ascend hardware. Illustration: DeAI
A datacenter rack aisle with accelerator cabinets and an amber status LED, illustrating Meituan's LongCat-2.5 open-weights model reportedly trained on Huawei Ascend hardware. Illustration: DeAI

Today in DeAI: Meituan shipped LongCat-2.5 as MIT-licensed open weights, OpenAI's usage-policy rewrite takes effect November 20, and GPT-5.6 Sol's promotional API pricing floor expires November 21.

Meituan ships LongCat-2.5 open weights under MIT

Meituan's LongCat team released the weights of LongCat-2.5, a 1.6-trillion-parameter multimodal mixture-of-experts model for long-horizon agent work, under the MIT license. Meituan reports 88.4 on Terminal Bench v2.1 and says the model was trained on domestic Chinese accelerators. Why it matters: a frontier-scale agent model with permissive weights and a promotional API rate near €0.27/€1.07 per million tokens widens real self-host and multi-provider options. Full story: our coverage (LongCat blog).

OpenAI usage-policy rewrite lands November 20

OpenAI's usage-policies changelog sets an October 21 update live on November 20: clarified restrictions on using OpenAI models or outputs to train other models, and approval requirements for model-access intermediaries. Why it matters: teams distilling from the OpenAI API or routing it through aggregators should read the effective text and confirm their intermediaries' approvals before the 20th. Related context: xAI's default watermarking move.

GPT-5.6 Sol promo pricing expires November 21

OpenAI's developer pricing page guarantees GPT-5.6 Sol's promotional $4/$20 per million tokens only "at least through November 21, 2026," with no successor rate published; Beri.net's analysis argues reversion is +50% on output tokens because the August cut was asymmetric. Why it matters: pin current rates and check whether your router passes through the change before output-heavy workloads reprice. Background: our pricing coverage.

Enterprise defaults flip midweek: Dots and Claude Data

Two enterprise defaults flip within 24 hours, according to a November roundup: OpenAI's Dots personal agents go on by default in ChatGPT Enterprise on November 18, and Anthropic's Claude Data turns on by default in Claude Enterprise on November 19. Both dates trace to vendor launch materials and remain unconfirmed in primary admin documentation, so treat them as claims until checked. Why it matters: admins running compliance-sensitive workloads have an opt-out window with a clock on it.

Solana Confidential Compute enters devnet beta

The Solana Foundation and Nito shipped Solana Confidential Compute, an open-source SDK and onchain program verifying Intel TDX and AMD SEV-SNP attestations for private AI workloads, in devnet beta. Why it matters: TEE-backed verifiable inference keeps moving onto mainstream rails. Full story: our PULSE coverage.

Watching tomorrow

The November 18 ChatGPT Enterprise Dots default-on is the next dated event to watch, followed by the Claude Data flip on November 19 and the OpenAI usage-policy effective date on November 20.

Sources

  1. Introducing LongCat-2.5 — LongCat (Meituan)
  2. meituan-longcat/LongCat-2.5 model card — Hugging Face
  3. OpenAI Usage Policies — OpenAI
  4. OpenAI GPT-5.6 Sol promotional pricing expiry analysis — Beri.net
  5. OpenAI developer changelog — OpenAI

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.

Learn more about the Morpheus Inference API →

Sponsor disclosure — not editorial

Powered by Morpheus and StrandCMS. Morpheus is a decentralized inference marketplace, covered on the same terms as every other provider. StrandCMS is the open-source, agent-first framework this site is built on.

Learn more →