model release

ricursiveAI's 65B Ripple Preview launches on GMICloud: 16K context, $0.20/$0.60 per 1M tokens

TL;DR

Ripple-65B-Preview, a 65B-parameter model from ricursiveAI with native tool calling, is served in BF16 by GMICloud and listed on OpenRouter as of Nov 4, 2026. It has a 16K context window and costs $0.20 per 1M input tokens and $0.60 per 1M output tokens.

1 min read
0

Ripple-65B-Preview at a glance

Context window1049K tokens
Input$0.1/1M tokens
Output$0.3/1M tokens

Ripple-65B-Preview, a 65B-parameter model from ricursiveAI with native tool calling, is now available through GMICloud. It is listed on OpenRouter under the ID gmicloud/ripple-65b-preview with a release date of Nov 4, 2026.

Specifications and pricing

  • Parameters: 65B
  • Precision: BF16
  • Context window: 16K
  • Input price: $0.20 per 1M tokens
  • Output price: $0.60 per 1M tokens
  • Cache read price: $0.05 per 1M tokens
  • Tool use: native tool calling
  • Status: Preview

GMICloud is the only provider listed for the model on OpenRouter.

Serving metrics (read Nov 4, 2026)

OpenRouter's listing shows the following for GMICloud:

  • Latency: 0.56s (P50, best provider)
  • Throughput: 89 tok/s (P50, best across providers)
  • Uptime: 100.00%

The uptime panel for the 3-day window reads "Not enough uptime data to display yet," and no app usage data is shown for the model yet. These are live statistics from launch day and will change.

Availability

The model is not currently available through EU in-region routing. OpenRouter notes that if an upstream provider errors, it can reroute to another healthy provider when request filters allow. With only GMICloud listed, there is no alternative provider on this page.

What this means

The 16K context window limits Ripple-65B-Preview to shorter prompts and conversations. Native tool calling makes it a candidate for function-calling and agent-style workflows within that limit. The Preview label signals that the model is not a final release. Teams with EU in-region routing requirements cannot currently use it that way.

Comments

Loading...