SSyncDev

REKI-2B

SyncDev model directory · Profile updated Nov 18, 2026

Compare with another model

REKI-2B is Kyutai's open 2.4B robot vision-language-action model with cheap register-token memory, pretrained only on public data.

Key specifications

Specification score
9.8
License
CC-BY-NC-SA-4.0 (weights); MIT (code)

The specification score summarizes recorded features and limits. It is not a benchmark result.

Pricing conditions: Open weights; no per-token price is published. Your cost is the GPU time you run it on.

Inputs and outputs

Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.

Accepts

  • Image input
  • Text input

Produces

  • Text output

Capabilities

VisionOpen weights

What is REKI-2B?

Robot policies without memory oscillate. Reach for the cup, forget why, reach back. REKI-2B, from Kyutai with Inria and École des Ponts, adds memory almost for free. Thirty-two register tokens in the action expert summarise what the vision-language model saw at each frame and persist in the expert's cache across the last three frames, one second apart. Reading them adds under 1% to the time it takes to predict a chunk.

The payoff, by Kyutai's measures: oscillations between poses become about four times rarer, and success rises 1.5 points on LIBERO-Plus, 5.5 on RoboTwin clean scenes and 5.4 on RoboCasa365.

Built from public data

Everything in pretraining is public: real-robot teleoperation (Open X-Embodiment subset, DROID, BridgeData V2, AgiBot World and more), simulation, egocentric human-hand video and vision-language data. One billion frames in all, plus 12.2M vision-language samples, across 14 state layouts from single arms to humanoids. Training ran 200,000 steps on 32 H100s.

Architecture

  • 2.4B parameters: a 2.2B Qwen3.5-2B vision-language model with 448px images, and a 0.2B flow-matching action expert.
  • The expert reads the final VLM layer through cross-attention.
  • Each embodiment gets its own state MLP feeding adaptive RMSNorm.
  • Compiled, a chunk takes 21.8 ms on one H100 from two 256 × 256 images.

Kyutai reports the best scores among public-data models on RoboTwin Clean2Rand (65.8) and LIBERO-Plus (90.4).

Releases and licence

The base appeared on 5 October 2026, with fine-tunes for LIBERO, RoboTwin (clean, and clean plus randomised) and RoboCasa365. Weights are CC BY-NC-SA 4.0, code is MIT. That rules out commercial deployment of the weights.

Real-robot use goes through LeRobot's async-inference gRPC protocol.

Getting a first result

Start from the fine-tune closest to your hardware, say the LIBERO Franka model for a Panda arm. Reproduce its benchmark numbers in simulation before touching a real robot. Then fine-tune on fifty of your own demonstrations and count oscillation events per episode with memory on and off. That is the claim worth checking.

For a commercial-licence-friendlier alternative, look at FLUX 3 Action. Kyutai's speech models are on its provider page.

How to evaluate REKI-2B

Compare the model on the work you need to complete.

Tasks and inputs

REKI-2B is listed for Image-text-to-text, Multimodal. It accepts image, text and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.

Capacity and cost

Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.

Choosing a deployment

The weights are downloadable under CC-BY-NC-SA-4.0 (weights); MIT (code). Check runtime support and size the complete model and working context for your expected concurrency.

Frequently asked questions

A 2.4B vision-language-action model for robots, with register-token memory, from Kyutai, Inria and École des Ponts.

These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.

NVIDIA

Alpamayo 2 Edge

Alpamayo 2 Edge is NVIDIA's open 2B-class driving VLA that outputs trajectories with causal explanations, sized for DRIVE AGX Thor in-vehicle use.

— context

shanghai-ai-lab

Sage-SciPerception

Sage-SciPerception is Shanghai AI Lab's set of projectors feeding crystals, spectra, biological sequences and biosignals into a language model's input.

— context

inclusionAI

Ling 3.2 Flash

Ling 3.2 Flash is inclusionAI's 560B-A25B multimodal MoE with a 1M-token window, built for visual agents, 3D tasks and medical image reasoning.

1M context

Swiss AI

Apertus 1.5 8B

Apertus 1.5 8B is the Swiss AI Initiative's open multilingual model, now taking text, images and audio with a 262K-token context window.

262.14K context

black-forest-labs

FLUX 3 Action

FLUX 3 Action is Black Forest Labs' open 7B world action model for robots, predicting action chunks and future video from cameras, state and text.

— context