feat(astro-orbiter): bump Qwen3.8-27B ctx-size 32768->131072 (128K) [t_441470b9]

ansible/roles/llm-inference-multimodel/defaults/main.yml

@@ -159,10 +159,12 @@ llm_router_phi_flash_attn: "{{ llm_router_flash_attn }}"
 llm_router_coder_ctx_size: 16384
 llm_router_coder_flash_attn: "true"
 llm_router_preset_path: /opt/llama-server-router-preset.ini
-# Qwen3.8-27B: ctx=32768 (32K). Measured VRAM: 17,068 MiB at 32K vs 17,804 MiB at 64K.
-# Using 32K to leave more headroom during LRU eviction transitions on the 24GB RTX 3090.
-# Native context of Qwen3.8-27B is 262,144 tokens; 32K is sufficient for Hermes.
-llm_router_qwen38_ctx_size: 32768
+# Qwen3.8-27B: ctx=131072 (128K). Bumped from 32768 -> 131072 per Ryan approval (t_441470b9, 2026-08-16).
+# Measured VRAM: 20,282 MiB at 131072 ctx (empirically tested in t_4455a44c); nomic-embed 558 MiB
+# always resident -> ~20.8GB total, ~3.2GB headroom on 24GB RTX 3090. Comfortably safe.
+# Prior value was 32768 (17,068 MiB) — bumping 4x for genuine 128K context.
+# Native context of Qwen3.8-27B is 262,144 tokens; 128K is a practical production ceiling.
+llm_router_qwen38_ctx_size: 131072
 # nomic-embed-text-v1.5: embedding model, ctx-size=8192 per task t_34b96e83
 # No flash_attn or KV cache params - embedding models use bidirectional forward pass,
 # not autoregressive KV cache. load-on-startup=true / sleep-idle-seconds=-1 keep it

ansible/roles/llm-inference-multimodel/templates/llama-server-router-preset.ini.j2

@@ -58,10 +58,11 @@
 ; KV cache at ctx=65536 (q4_0 KV). Smaller than prior Qwen3.6 at ~20GB.
 ; Native context: 262,144 tokens. Running at 65536 (Hermes floor) for now;
 ; can be raised later if needed.
-; VRAM footprint (measured 2026-08-16): 17,068 MiB at ctx=32768 with q4_0 KV;
-;   17,804 MiB at ctx=65536. Using 32768 (32K) to give more eviction headroom
-;   on the 24GB RTX 3090 (nomic-embed 558MB always resident; total ~17.6GB).
-;   Native context is 262,144 tokens; 32K is sufficient for Hermes usage.
+; VRAM footprint (empirically tested, t_4455a44c 2026-08-16):
+;   ctx=32768:  17,068 MiB; ctx=65536: 17,804 MiB; ctx=131072: 20,282 MiB.
+;   BUMPED to 131072 (128K) per Ryan approval (t_441470b9, 2026-08-16).
+;   nomic-embed always resident at 558 MiB -> total ~20.8GB, ~3.2GB headroom.
+;   Native context is 262,144 tokens; 128K is the production ceiling.
 ; Architecture note: Qwen3.8 uses Gated DeltaNet; llama.cpp 6ea215d logs
 ; "fused Gated Delta Net (chunked) not supported, set to disabled" — falls
 ; back to non-fused implementation. Inference works correctly but may be
@@ -73,6 +74,7 @@
 ; them in turn. One cold-load (~30-60s) per switch between Qwen3.8 and
 ; auxiliary models is expected and acceptable. In practice, once Hermes
 ; config.yaml references Qwen3.8 as primary, it stays resident.
+; ctx-size raised to 131072 (128K) per Ryan approval (t_441470b9, 2026-08-16).
 ; Primary model ID: Qwen3.8-27B-Q4_K_M
 [Qwen3.8-27B-Q4_K_M]
 model              = {{ llm_models_dir }}/Qwen3.8-27B-Q4_K_M.gguf