Add Darwin-36B-Opus VKAE row (11.2x, 25->280.8) + links

Leaderboard: add Links column (Docker + HF)

Add docker/HF links to Qwen35B leaderboard row

Add JGOS-398B measured row (4.33x, 88->382) to leaderboard series

Remove Qwen3-32B benchmark entry (1.1x)

Remove low-gain Qwen3-32B row (1.1x) from leaderboard

Redact kernel recipe -> optimized serving in benchmark rows

Redact residual method terms (kernel recipe / vLLM/SGLang) to generic serving

Redact residual method terms (kernel recipe / vLLM/SGLang) to generic serving

Qwen3.5-35B-A3B series -> 601 tok/s (23.4x, peak 10,516); redact method hints

Add VKAE Acceleration Leaderboard (ranked sortable table, method withheld)

Update README.md

Qwen3.5-35B-A3B: show 601 tok/s (23.4x vs reference), varied ~455 + peak 10,516 in role; method withheld

Update Qwen3.5-35B-A3B row: 25.7->455 tok/s (17.7x, FP8+optimized), peak 10,516 tok/s; method withheld

Redact acceleration-method disclosure (graph/vLLM/SDPA) from public rows; keep claim-scoped generic labels

Update README.md

add Qwen3.5/3.6 measured VKAE graph rows

add Qwen3.5-9B(3.41x) Qwen3.6-27B(2.47x) Qwen3.5-35B-A3B MoE(10.35x) measured rows

add Qwen3-8B/32B measured VKAE graph rows

add Qwen3-8B (1.85x) + Qwen3-32B (1.10x) VKAE CUDA-graph measured rows

remove 12B/26B/31B rows; keep E4B showcase + 398B flagship

remove Gemma 12B/26B-A4B/31B (gemma4_unified, unservable in vLLM, ~1x preview)

remove Gemma E2B row (package-validation only, 1.06x)

remove Gemma E2B entirely (E2B real speedup 1.06x, not a VKAE showcase)

revert E2B: unverified speedup removed (serve logs show ~1x, not 4.1x)

E2B: fill measured B200 VKAE graph numbers (82.9->340.7, 4.1x)

Add JGOS-398B (Qwen3.5-MoE) large-MoE serving-acceleration lane: 4.33x (88->382 TPS, B200x6 TP2PP3); method internals withheld

Update VKAE productization plan

Add VKAE OpenAI compatibility surface

Gate VKAE before-after claims

Update VKAE Gemma model packages and preview benchmark metadata

Publish VKAE-Gemma-E2B preview bundle

Add VKAE-Gemma-E2B package status

Remove rejected candidate from public VKAE metrics

Fix VKAE claim-scoped benchmark presentation

Mark VKAE candidate confirm16 gate

Update VKAE repeat-2 package benchmark

Productize VKAE-Gemma E4B private preview

Separate optimized VKAE claims from preview measurements

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_complete

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Upload space_vkae_gemma/app.py with huggingface_hub

Load benchmark results from latest Hub artifact

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Update VKAE H200 BF16 fallback benchmark results: h200_bf16_running

Mark H200 fallback benchmark job