Unverified50% confidenceFactExact time
The text encoder of LTX 2.5 uses the Gemma 4 model with 12 billion parameters released by Google DeepMind.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/10/2026
First Seen
Valid until: 12/9/2026
Sources
Related Entities
Related Claims
UnverifiedGoogle DeepMind released Gemma 4 quantization-aware training weights, reducing the memory footprint of the 12B parameter model to approximately 1GB through targeted 2-bit compression technology.72% similarVerifiedGoogle DeepMind's FunSearch has used LLMs to discover new knowledge in the field of mathematics.69% similarUnverifiedGoogle DeepMind introduced MJX (MuJoCo XLA) and Brax, both based on JAX, compiling physics simulation into computation graphs that can be efficiently executed on GPUs/TPUs.68% similarVerifiedGoogle DeepMind's AlphaProof combines a large language model with the formal proof verifier Lean, adopting a 'generation-verification' closed-loop mechanism.67% similarUnverifiedGoogle DeepMind的RT-2（2023年）将连续的末端执行器位姿和夹爪开合量量化为256个离散区间，附加到视觉-语言模型PaLI-X和PaLM-E的输出词汇表中67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/890279API
curl https://kongchang.com/api/v1/knowledge/claims/890279MCP
get_claim(id=890279)