Orkun Soylu
  1. Strata @W7800 + RX 7900 XTX
  2. Strata 125B MoE @48 GB vs Qwen3.8-27B
  3. Coder390 EfficientThink vs Qwen3.8-27B
  4. DavidAU TURBO & Swift vs Qwen3.8-27B: düşünme token'ları
  5. llama.cpp ROCm vs Vulkan @W7800
  6. W7800: fiziksel makine vs Proxmox VM
  7. Kendi Q6_K'm (Türkçe imatrix) vs Q8_0 vs ISTA IQ3_S
  8. MTP spekülatif çözme @Ollama: 20 → 42 tok/s
  9. empero 35B-A3B vs Qwen3.8-27B Q8+MTP vs ISTA IQ3_S @CVE triyajı
  10. vLLM vs Ollama @W7800 (ROCm)
  11. ornith-1.5 35B MoE @CVE triyajı
  12. ISTA IQ3_S: genel yetenek
  13. HauhauCS Aggressive Q8 vs 3-bit @CVE triyajı
  14. CVE sistem prompt'u v2: ISTA IQ3_S @0,98
  15. CVE sistem prompt'u v1
  16. ISTA IQ3_S vs Unsloth UD-IQ3_S @16 GB
  17. ISTA IQ3_S vs HauhauCS IQ3_XS @16 GB: özellikler
  18. ISTA IQ3_S vs HauhauCS IQ3_XS @CVE triyajı
  19. vLLM vs Ollama @RTX 4070 Ti SUPER
  20. Qwen3.8-27B: düşünme açık vs kapalı
  21. pi harness: Qwen3.8-27B Q4_K_M vs Q8_0
  22. Dense 27B @168k token prompt
  23. W7800 vs RX 7900 XTX
  24. Qwen3.8-27B Q8_0 vs Q4_K_M
  25. 48 GB'a sığan en uzun bağlam: beş model
  26. Qwen3.6 35B-A3B MoE vs 27B @26k token prompt