Unverified50% confidenceFactExact time
Inception's Mercury 3 uses a diffusion architecture, filling multiple tokens at once, reaching 1,068 tokens per second on Nvidia GPUs, and has a 1 million token context window.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/20/2026
First Seen
Valid until: 1/18/2027
Sources
Devin Now Commits 89% of Cognition's Code as Company Hits $1B Annualized Revenue
youtubeNerra Network10/20/2026
Related Entities
Related Claims
UnverifiedInception Labs released Mercury 3, a diffusion LLM, claiming it surpasses GPT-6 Luna and Claude Haiku 5.5 on benchmarks and runs 4 to 8 times faster; generation speed reaches 1,068 tokens/second on NVIDIA GPUs, and the context window expands from 260K to 1 million tokens.74% similarUnverifiedNVIDIA's validated FP4 version of Qwen3.8 configuration offloads the n-gram embedding table to the CPU (VLLM_PLE_CPU_OFFLOAD=1), freeing up 850K tokens of FP8 KV cache.65% similarUnverifiedMercury 3 is available only through Inception's Nvidia GPU API, with no open weights and no quantized versions.63% similarVerifiedNVIDIA introduced Tensor Cores starting with the Volta architecture, which are hardware units specifically designed for matrix multiply-accumulate operations.61% similarUnverifiedTokenSpeed requires an NVIDIA GPU host (the nightly wheel requires CUDA 13) or an AMD host with ROCm 7.2, and Python 3.11 to 3.13.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074811API
curl https://kongchang.com/api/v1/knowledge/claims/1074811MCP
get_claim(id=1074811)