Ricursive Intelligence · llm model

Ripple-65B Preview

Ripple-65B Preview is Ricursive Intelligence's latest language model in our dataset (public preview), released Nov 4, 2026. Open weights (Apache 2.0). 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. Verified against vendor sources on Nov 4, 2026.

Reported Last verified Nov 4, 2026 · All models · JSON · Markdown

Primary source: https://www.ricursive.com/ripple. We are the fast, independent index; the vendor’s own page is the source of record for these model facts.

Ripple-65B Preview at a glance: 1,000,000-token context window; takes text in.

ReleasedNov 4, 2026
Context window1M1,000,000 tokens
LicenceOpen weightsApache 2.0

InTextOutnot verified

Key facts

Vendor-reported facts are read from the linked vendor page on the date shown above; we did not run the model. Independently tested means we ran it; no fact on this page is. Editorial is our own classification, not a vendor claim. Where a linked page does not itself state a value, the row says so. How to read provenance.

Ripple-65B Preview: key facts with sources
Fact	Value	Label	Source
Vendor	Ricursive Intelligence	Editorial	n/a
Released	Nov 4, 2026 Announcement date; the Hugging Face repository went public the same day.	Vendor-reported	source
Status	public preview	Editorial	n/a
Context window	1,000,000 tokens (1M) The vendor says "up to 1M context length"; pretraining used 8K sequences and later long-context stages ran at 2M, per the model card.	Vendor-reported	source
Input	text	Vendor-reported	source
Parameters	65B total, dense Transformer	Vendor-reported	source
Licence	Apache 2.0 (open weights)	Vendor-reported	source

Where Ripple-65B Preview sits in our data

Released Nov 4, 2026 (week 45 of 2026), in November 2026. Other models in our data released that month: Hy4, Sarvam Prism, Hydrogen-1.0, LongCat-2.5.
Of the 69 models in our data with a published context window, Ripple-65B Preview has the number 21 largest (1M tokens). Our data is not every model on the market.
All releases by date: AI model releases by month · new AI models this week. Choosing for code: best LLM for coding. Price and rank context: LLM pricing comparison · cheapest LLM API · OpenRouter pricing · best AI model right now.

Alternatives to Ripple-65B Preview

Language models from other vendors that share at least one of our groups with Ripple-65B Preview. Comparable on paper, from each model’s sourced facts; we have not run them.

Sarvam PrismSarvam AI · Both in: Open weights · 33K context · released Nov 3, 2026
Hydrogen-1.0Deep Principle · Both in: Open weights · released Nov 2, 2026
Clover-35B-A3B-PreviewDiffuSpace · Both in: Open weights · released Oct 30, 2026
SenseNova-Loop-9B-PreviewSenseNova · Both in: Open weights · 262K context · released Oct 30, 2026
LFM3-1.7BLiquid AI · Both in: Open weights · 256K context · released Oct 27, 2026
Nemotron 3.5 Bolt E4B (4GB)NVIDIA · Both in: Open weights · 64K context · released Oct 23, 2026

Side by side: Best open-weight LLMs (Nov 2026): GLM-5.3 leads Terminal-Bench 4.0 · Longest context window LLMs

What the sources say

These are claims from the linked pages, not our measurements.

Ricursive reports 80.52 on Terminal-Bench 2.1 and 76.5 on SWE-bench Verified for the 65B dense model. Vendor claim; we did not run it.

Notes

The release includes a BF16 instruction-tuned model, an NVFP4 checkpoint and a BF16 base model; Ricursive says the model fits on an NVIDIA DGX Spark.

What we tested

Nothing yet. We have not run Ripple-65B Preview ourselves, so every fact on this page is Reported from the sources below. When we do test it, the scope will be stated here and in our test logs. See methodology.

Not verified, left out

We could not confirm these on a vendor page, so we do not state them.

Hosted pricing on Together AI, GMI Cloud, OpenRouter and Fireworks was not checked, so no price is listed.
Benchmark scores are Ricursive's own; some comparison scores in its table come from its in-house eval flow, so they are listed as vendor claims only.