Expired85% confidenceFactTime unknown
The APEX quantized version is described as 'Q4 size, Q8 experience' and achieved the same score as the 35B NVFP4 in the CLI40 test.
1
Sources
85%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedIn the Q4_K_XL scheme, MoE layers use MXFP4 and the remaining layers are compressed to Q8_0, achieving 96% same on top-1% prediction consistency64% similarUnverifiedQwen3.6-27B's benchmark scores place it at the leading position among models of similar scale (20B–30B parameter range).63% similarUnverifiedQwen3.6 Plus (API version) scored only 76 in the command-line operations (CLI40) test.63% similarUnverifiedThe OPUS 4.6 distilled version comes from the Jacker community and performed well in the Qwen 3.5 era, but the 3.6 V1 preview test results were unsatisfactory, and its Q6 precision also falls short of NVFP4.62% similarUnverifiedThe original Qwen3.6 27B Q4 scored 97 in the tool-calling test, demonstrating that distillation does not always improve performance.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/17963API
curl https://kongchang.com/api/v1/knowledge/claims/17963MCP
get_claim(id=17963)