{"data":[{"id":"4b48c472-c4e1-559c-923c-b6f116af0856","title":"InferenceX benchmark result 93094","sourceType":"benchmark","url":"https://inferencex.semianalysis.com/api/v1/benchmarks?model=DeepSeek-R1-0528&runId=21407590933&exactRun=true","repository":"sgl-project/sglang","observedAt":"2026-01-27T22:56:48.000Z","sanitizedSha256":"09f1a7084285486fed350da4d22d7af2c4c5d853cd89a9a596a651187fce6888","freshness":"stale","note":"The public InferenceX API reports 92.94321217652133 output tokens per second per GPU for DeepSeek R1 on eight MI355X GPUs at concurrency four, with 1,000 input and 1,000 output tokens."},{"id":"abb7d4b9-c6a2-50ce-99a6-db4477df7b10","title":"InferenceX run 21407590933 attempt 2","sourceType":"ci_attempt","url":"https://github.com/SemiAnalysisAI/InferenceX/actions/runs/21407590933/attempts/2","repository":"sgl-project/sglang","observedAt":"2026-01-27T22:56:48.000Z","sanitizedSha256":"32ba269d2b318c51936ea8321621e94548d5b9f720bc4cf235421d7b01b6eaae","freshness":"stale","note":"The benchmark workflow completed successfully for InferenceX merge commit 2a796d10eca68071e5830c2d1a9a00287cbf51c0."},{"id":"28b962ac-9907-51e7-9a39-18489c93d12b","title":"InferenceX PR 572 adds SGLang v0.5.8 MI355X results","sourceType":"pull_request","url":"https://github.com/SemiAnalysisAI/InferenceX/pull/572","repository":"sgl-project/sglang","observedAt":"2026-01-27T17:35:59.000Z","sanitizedSha256":"dff011e3b2e496aa199ac9ef5b79b3eddca1bd16908af081dfd20aa31d86b258","freshness":"stale","note":"The InferenceX changelog links SGLang PR 17327 and the dsr1-fp8-mi355x-sglang configuration."},{"id":"abf24710-c99c-53dc-872b-2b75a1fd5678","title":"InferenceX performance changelog linkage","sourceType":"benchmark","url":"https://github.com/SemiAnalysisAI/InferenceX/blob/f50e14b0db83c2ca685c6acd701cf290b1d35710/perf-changelog.yaml","repository":"sgl-project/sglang","observedAt":"2026-01-27T17:35:59.000Z","sanitizedSha256":"b8cabcec722d790c4bf2dac711394c8aeaae733413ca348f97dee7b962bfe832","freshness":"stale","note":"The public changelog ties the benchmark update to SGLang PR 17327 and the v0.5.8 configuration."},{"id":"d411351f-0e75-5b69-b00c-37a0dad74ac0","title":"InferenceX SGLang MI355X benchmark configuration","sourceType":"benchmark","url":"https://github.com/SemiAnalysisAI/InferenceX/blob/f50e14b0db83c2ca685c6acd701cf290b1d35710/.github%2Fconfigs%2Famd-master.yaml","repository":"sgl-project/sglang","observedAt":"2026-01-27T17:35:59.000Z","sanitizedSha256":"eaeaf509683d968b1cb2feea3e70cfc6ec343af7ca5da6e4a5da021c17f5d91a","freshness":"stale","note":"The configuration pins image lmsysorg/sglang:v0.5.8-rocm700-mi35x for the DeepSeek R1 FP8 MI355X run."},{"id":"ee9c8936-8295-54bb-9455-2622c45e31e5","title":"Release v0.5.8","sourceType":"release","url":"https://github.com/sgl-project/sglang/releases/tag/v0.5.8","repository":"sgl-project/sglang","observedAt":"2026-01-25T00:40:11.000Z","sanitizedSha256":"d2d1d4982279964c7f81754028bd57717ea6883ed4040a24d8abfe7d17db3612","freshness":"stale","note":"Release observed"},{"id":"16696ec6-9cc5-5ac9-8e3c-c6f20b1bffa4","title":"SGLang v0.5.8","sourceType":"release","url":"https://github.com/sgl-project/sglang/releases/tag/v0.5.8","repository":"sgl-project/sglang","observedAt":"2026-01-23T22:09:28.000Z","sanitizedSha256":"1cd9b9bdb52c4d29e3d53e78243284fc2301ff1654c0a56723d458a30e857ff1","freshness":"stale","note":"SGLang published v0.5.8 from commit 0189f41c30ede088a040a711a384f3024b8d7af5."},{"id":"4154e592-764f-5b45-9f1b-9aa508d265b5","title":"SGLang v0.5.8 contains PR 17327 merge commit","sourceType":"commit","url":"https://github.com/sgl-project/sglang/compare/6988a0f5706c05ebc40ba9ef5c09243bcbf9886c...0189f41c30ede088a040a711a384f3024b8d7af5","repository":"sgl-project/sglang","observedAt":"2026-01-23T22:09:28.000Z","sanitizedSha256":"0542e7a91617de186255aa2de36a51c0815d2b1835d62417738a88e4742786c3","freshness":"stale","note":"GitHub compare reports the v0.5.8 commit is 73 commits ahead and zero commits behind merge commit 6988a0f5706c05ebc40ba9ef5c09243bcbf9886c."},{"id":"3e2084f6-6026-567c-a82a-5655232dda72","title":"SGLang PR 17327 disables the persistent MLA kernel without FP8 KV cache","sourceType":"pull_request","url":"https://github.com/sgl-project/sglang/pull/17327","repository":"sgl-project/sglang","observedAt":"2026-01-20T04:29:54.000Z","sanitizedSha256":"95155a85c57c66b0935e33dd53407de965e70a4cbe822c0000dde3f3d75b9f01","freshness":"stale","note":"Merged SGLang change 6988a0f5706c05ebc40ba9ef5c09243bcbf9886c updates the ROCm AITER attention backend guard."},{"id":"c28a0d98-2160-5ddd-ab9c-3b900f3f53ff","title":"SGLang AMD DeepSeek R1 MXFP4 eight-GPU test definition","sourceType":"test","url":"https://github.com/sgl-project/sglang/blob/35443039efe9c82aa73af6e8692fe7b2c5bd308d/test/registered/amd/test_deepseek_r1_mxfp4_8gpu.py","repository":"sgl-project/sglang","observedAt":"2026-01-20T04:29:54.000Z","sanitizedSha256":"b6de7bfce3039aef3973cf12032f0cc705dba93782d10a5e8945273a8636effd","freshness":"stale","note":"The registered AMD test requires GSM8K accuracy above 0.94 and speed above 75 output tokens per second."},{"id":"f8b42383-23a4-5330-a61d-428bba038648","title":"SGLang AMD pull request workflow","sourceType":"workflow","url":"https://github.com/sgl-project/sglang/blob/35443039efe9c82aa73af6e8692fe7b2c5bd308d/.github/workflows/pr-test-amd.yml","repository":"sgl-project/sglang","observedAt":"2026-01-20T04:29:54.000Z","sanitizedSha256":"b8b04df86423d0afae58c4281f831e473e3db4b363f410047d36382924bfd5c3","freshness":"stale","note":"The PR workflow runs the registered AMD test suite in three automatically selected partitions."},{"id":"427adf3c-9611-5587-bd96-2a1f66e60998","title":"SGLang PR Test AMD run 21151393674 attempt 2","sourceType":"ci_attempt","url":"https://github.com/sgl-project/sglang/actions/runs/21151393674/attempts/2","repository":"sgl-project/sglang","observedAt":"2026-01-20T03:56:34.000Z","sanitizedSha256":"0d700cb8a0f3462740d0970475cc3ded8b82d685bfd6e6bb6ed234e685eaef14","freshness":"stale","note":"GitHub Actions reports attempt 2 completed successfully for PR head 35443039efe9c82aa73af6e8692fe7b2c5bd308d."},{"id":"1996ca26-1b36-55ae-80ee-5da5b109d0af","title":"SGLang AMD suite partitions for run 21151393674 attempt 2","sourceType":"test","url":"https://github.com/sgl-project/sglang/actions/runs/21151393674/attempts/2","repository":"sgl-project/sglang","observedAt":"2026-01-20T03:56:26.000Z","sanitizedSha256":"e1ca050740f74d976c6c0b0f231721d93459a199e1d416384ead82e8b0b1c1b9","freshness":"stale","note":"GitHub reports success for AMD test partitions 0, 1, and 2. Retained artifacts are unavailable, so this evidence does not claim an exact file partition or assertion value."}],"page":{"nextCursor":null,"hasMore":false,"limit":50},"meta":{"asOf":"2026-08-17T16:37:40.258Z","snapshotId":"weekly-cutoff:pilot-live-v1:2026-08-17T16:00:00.000Z","sourceCutoff":"2026-08-17T16:00:00.000Z","freshness":"unavailable","taxonomyVersion":"513f67b1-9ea7-46cc-b713-d6ea30ffb93f","metricVersions":{"ci":"ci-health:v2","contribution":"contribution-health:v2"},"freshnessPolicyVersion":"freshness-v1","cohortVersion":"aac24f2b-1109-5deb-a608-1c3d55fb0ae6","queryFingerprint":"ee0ba8c2880b9628a7e5f922e70a14542b29d5dd9a3e74e8f9b24372f91d61ab","sourceMode":"postgres","apiVersion":"v1"}}