Compare / Ripple 65B Preview vs Kimi Linear 48B-A3B
Ripple 65B Preview vs Kimi Linear 48B-A3B
Updated November 2026
Verdict
Ripple 65B Preview scores higher on the Small LM Index: 74.6 vs 36.2. On the 5 benchmarks both labs report, Ripple 65B Preview wins 5 and Kimi Linear 48B-A3B wins 0. Kimi Linear 48B-A3B is smaller: it needs about 30.8 GB of memory at 4-bit, against 41.9 GB for Ripple 65B Preview. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Ripple 65B Preview if…
- you want the strongest model overall
- you need math and reasoning
- you need precise instruction following
- you work with long documents
Pick Kimi Linear 48B-A3B if…
- you have limited memory (about 30.8 GB at 4-bit)
- you want the more widely used model (more quantized builds and community support)
Specifications
| Ripple 65B Preview | Kimi Linear 48B-A3B | |
|---|---|---|
| Small LM Index | 74.6 (likely 74–77) | 36.2 (likely 33–39) |
| Lab | Ricursive Intelligence · Other | Moonshot AI (Kimi) · China |
| Released | 2026-11-04 | 2025-10-30 |
| Parameters | 64.8B (dense) | 48B (3B active, MoE) |
| Memory at 4-bit | ~41.9 GB | ~30.8 GB |
| Context window | 1M tokens | 1M tokens |
| Input | text | text |
| Reasoning | Always thinks | No thinking |
| License | Apache-2.0 | MIT |
| API price / 1M tokens | Self-host only | Self-host only |
| Hugging Face downloads (30 days) | 15K | 45.4K |
Benchmarks
Higher is better. Bold marks the winner where both labs report a score. ≈ marks the index model's estimate for a score that wasn't reported, ✓ an independent measurement and * a score measured by another lab.
| Benchmark | Ripple 65B Preview | Kimi Linear 48B-A3B |
|---|---|---|
| MMLU-Pro Knowledge | 84.2 | 58.5✓ |
| GPQA Diamond Knowledge | 85.2 | 41.2✓ |
| Humanity's Last Exam Reasoning | ≈ 18 | 2.5✓ |
| AIME Math | 98.3 | 36.3✓ |
| LiveCodeBench Coding | ≈ 90 | 37.8✓ |
| SWE-bench Verified Coding | 76.5 | ≈ 38 |
| Terminal-Bench Agentic | 80.5 | — |
| τ²-Bench Agentic | — | 0.0✓ |
| IFBench Instruction following | 79.2 | 28.1✓ |
| MMLU Knowledge | 89.6 | — |
| HMMT Math | 90.5 | — |
| AA-LCR Long context | 75.0 | 28.0✓ |
| SciCode Coding | 39.4 | — |
Scores are mostly reported by each lab with its own settings, so small gaps are noise. See the methodology. Open in the interactive comparison to add more models.
Questions
Which is better, Ripple 65B Preview or Kimi Linear 48B-A3B?
Ripple 65B Preview scores higher on the Small LM Index: 74.6 vs 36.2. On the 5 benchmarks both labs report, Ripple 65B Preview wins 5 and Kimi Linear 48B-A3B wins 0.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Kimi Linear 48B-A3B is smaller: it needs about 30.8 GB of memory at 4-bit, against 41.9 GB for Ripple 65B Preview.
Which has the longer context window?
Ripple 65B Preview has the longer context window: 1M tokens against 1M.
Can I use Ripple 65B Preview and Kimi Linear 48B-A3B commercially?
Ripple 65B Preview is released under Apache-2.0 and Kimi Linear 48B-A3B under MIT. Apache 2.0 and MIT allow commercial use; other licenses can add conditions, so read the license before shipping a product.