AI model releases, November 2026
5 model releases in our dataset for November 2026, from Tencent, Ricursive Intelligence, Sarvam AI, Deep Principle, LongCat (Meituan). The newest is Hy4 (Nov 4, 2026).
Period Nov 1, 2026 to Nov 30, 2026 · last verified · Reported from vendor pages, not run by us · this lists releases we track, not every model a vendor shipped · Markdown
| Date | Model | Vendor | Status | Release date source |
|---|---|---|---|---|
| Hy4 | Tencent | generally available | source · Tencent's announcement is dated November 4, 2026; Vercel's AI Gateway changelog entry is dated November 3, 2026. | |
| Ripple-65B Preview | Ricursive Intelligence | public preview | source · Announcement date; the Hugging Face repository went public the same day. | |
| Sarvam Prism | Sarvam AI | generally available | source · Date on Sarvam's release post. | |
| Hydrogen-1.0 | Deep Principle | generally available | source · The announcement is undated; the Hugging Face repo was last modified 2026-11-02 09:00 UTC (created gated on 2026-10-31). | |
| LongCat-2.5 | LongCat (Meituan) | generally available | source · Change log entry "LongCat-2.5 Now Available", version 2026-11-02. The blog post is dated 2026-11-1. |
Why does each release matter?
- Hy4 (Tencent, ): Tencent's Nov 4, 2026 flagship Hy model, the first in the line with vision and computer use (CLI and GUI), on Tencent Cloud, OpenRouter and Vercel AI Gateway; Tencent plans to open-source the weights later. Our brief: Tencent releases Hy4 with vision and computer use, at $0.834 per million input tokens
- Ripple-65B Preview (Ricursive Intelligence, ): Ricursive Intelligence's first open-weight model, released Nov 4, 2026: a 65B dense Transformer trained from scratch on 12T tokens with up to 1M context, tuned for agentic coding. Our brief: Ripple-65B Preview: a dense 65B open-weight model that reports 80.5 on Terminal-Bench 2.1, Apache 2.0
- Sarvam Prism (Sarvam AI, ): Sarvam AI's open-weights decision model built on Sarvam 105B, released Nov 3, 2026: it returns calibrated probabilities for fixed answer options (routing, triage, moderation, judging agent runs) instead of generating text. Our brief: Sarvam Prism: an open-weights decision model on Sarvam 105B that returns probabilities, not text
- Hydrogen-1.0 (Deep Principle, ): Deep Principle's open-weight scientific model for chemistry, materials science and life sciences, post-trained from GLM-5.3 and released under MIT on Nov 2, 2026. Our brief: Hydrogen-1.0: Deep Principle's 753B open-weight science model, post-trained from GLM-5.3 under MIT
- LongCat-2.5 (LongCat (Meituan), ): LongCat's native multimodal agent and coding model, available on the LongCat API from Nov 2, 2026 with a 1M-token context and a limited-time price of $0.30 input and $1.20 output per 1M tokens. Our brief: LongCat-2.5: a 1.6T-parameter multimodal agent model, live on the LongCat API at $0.30/$1.20
What are the key facts for each release?
Hy4
Tencent ·
InTextImageOutText
Context window 1M
Ripple-65B Preview
Ricursive Intelligence ·
InTextOutnot verified
Context window 1M
Sarvam Prism
Sarvam AI ·
InTextOutnot verified
Context window 33K
Hydrogen-1.0
Deep Principle ·
InTextOutText
Context window: not verified
LongCat-2.5
LongCat (Meituan) ·
InTextImageOutText
Context window 1M