From 49ff151b048b80bcf19ecd386ea3e9afe8711928 Mon Sep 17 00:00:00 2001 From: "claude[bot]" <41898282+claude[bot]@users.noreply.github.com> Date: Tue, 27 Oct 2026 04:16:55 +0000 Subject: [PATCH 1/2] chore(dsv41flash): update H100 vLLM AgentX image to nightly-00b6c4c Co-Authored-By: Claude Opus 5.5 --- inferencex-e2e/configs/nvidia-master.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/inferencex-e2e/configs/nvidia-master.yaml b/inferencex-e2e/configs/nvidia-master.yaml index 9ead101629..fafe6ba804 100644 --- a/inferencex-e2e/configs/nvidia-master.yaml +++ b/inferencex-e2e/configs/nvidia-master.yaml @@ -8667,7 +8667,7 @@ qwen3.5-fp8-gb300-dynamo-sglang-mtp: # caps batched tokens at 4096. Concurrency is bounded by the measured KV # ceiling rather than a guess; see the search-space comment. dsv41flash-fp4-h100-vllm-agentic-dspark: - image: vllm/vllm-openai:nightly-e4a5b5537344b103ed10de7764378ef09604ef3d@sha256:6b90105076216e861cd30fae8828efef9263a0399137b894289f3b6a685bf4ff + image: vllm/vllm-openai:nightly-00b6c4c565a8732c131e1f452d4a17803bb0842b@sha256:76a9614ed4ed20694914c87e82a0344bd73e1c6e081459a2467175ab903657e4 model: deepseek-ai/DeepSeek-V4.1-Flash model-prefix: dsv41flash runner: cluster:h100-dsxe From 5ac6f3a18b70b9a84fb308a2da67a10baa7c48b6 Mon Sep 17 00:00:00 2001 From: "claude[bot]" <41898282+claude[bot]@users.noreply.github.com> Date: Tue, 27 Oct 2026 05:51:51 +0000 Subject: [PATCH 2/2] chore(perf-changelog): add DSv4.1 Flash H100 vLLM AgentX image refresh Co-Authored-By: Claude Opus 5.5 --- inferencex-e2e/perf-changelog.yaml | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/inferencex-e2e/perf-changelog.yaml b/inferencex-e2e/perf-changelog.yaml index fd894cc02f..caabb6d187 100644 --- a/inferencex-e2e/perf-changelog.yaml +++ b/inferencex-e2e/perf-changelog.yaml @@ -11413,3 +11413,10 @@ - Restore the MI355X MiniMax-M3 FP4 vLLM AgentX TP2/TP4 grid from run 34609443618 attempt 2 using SRT recipes. Update nightly-2a02f6efe319c885e3ccbcecde402e0028f9ec1e to nightly-df6603f0583375f244d1ac6d36265462b7490392 (2026-10-22, digest sha256:81899892d73777a4ae81e16f39e749169687f29249505eb4c56da83bd097d41c). - Enable upstream MiniMax-M3 monokernel at TP4 with Model Runner V2, AITER MoE, FP8 KV/indexer and block size 128; keep TP2 on ordinary decode. Preserve EAGLE3-GQA with three draft tokens and use automatic golden acceptance for throughput with real verification for evals. pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/4418 + +- config-keys: + - dsv41flash-fp4-h100-vllm-agentic-dspark + description: + - "Update the vLLM image from nightly-e4a5b5537344 to nightly-00b6c4c565a8, pinned by digest." + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/4569 + scenario-type: [agentic-coding]