Unverified50% confidenceFactExact time
Kolibri uses a Mixture of Experts (MoE) architecture with approximately 78 billion total parameters, activating only approximately 3.46 billion parameters per token, and is open under the Apache 2.0 license.
1
Sources
50%
Confidence
Long-term
Relevance
10/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAlibaba Qwen's next-generation model has 2.4 trillion parameters, making it the first multimodal large model in the series to surpass one trillion parameters, and is open-sourced under the Apache 2.0 license.68% similarUnverifiedAlibaba's Qwen3 has approximately 2.4 trillion parameters, is about to be released and open-sourced, and the full-weight Max preview version is already online.62% similarUnverifiedKolibri-1 supports a context length of up to 1 million tokens.60% similarVerifiedMeta's Llama 3 has up to 405 billion parameters.60% similarUnverifiedDeepSeek's flagship version has approximately 1.6 trillion total parameters and 49 billion activated parameters; the lightweight Flash version has 284 billion total parameters and 13 billion activated parameters, supports a context window of 1 million tokens, and is open-sourced under the MIT license.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/973387API
curl https://kongchang.com/api/v1/knowledge/claims/973387MCP
get_claim(id=973387)