https://community.lizard-llm.qendryx.com/topic/11/read-this-first-what-the-benchmark-actually-measures2026-07-11T10:50:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/14/start-here-what-lizard-is-and-how-to-get-going2026-07-11T10:50:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/3/what-lizard-native-is-and-when-lizard-picks-it2026-07-11T10:50:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/17/ask-anything-no-question-too-basic2026-07-12T10:33:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/7/what-caterpillar-is-and-why-there-are-two-engines2026-07-12T10:33:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/20/show-us-what-you-ve-made2026-07-13T10:16:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/4/why-the-first-load-is-slow-and-every-one-after-is-fast2026-07-13T10:16:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/15/using-lizard-from-your-own-code-python-node-curl2026-07-14T10:59:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/8/speculative-decoding-why-repetitive-text-suddenly-runs-faster2026-07-14T10:59:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/12/post-your-numbers-shared-results-thread2026-07-15T10:42:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/18/which-model-should-i-pick2026-07-15T10:42:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/23/where-do-your-tokens-actually-come-from2026-07-15T11:18:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/5/running-a-model-that-s-bigger-than-your-graphics-card2026-07-16T10:25:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/21/offline-and-air-gapped-setups2026-07-17T10:08:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/9/how-many-cpu-threads-should-caterpillar-use2026-07-17T10:08:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/24/lizard-native-the-whole-inference-stack-running-locally2026-07-17T11:52:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/13/why-the-same-machine-gives-different-numbers-each-run2026-07-18T10:51:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/16/the-trial-credits-and-what-stays-yours2026-07-18T10:51:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/1/welcome-to-the-lizard-llm-community2026-07-18T14:26:47.771Zdaily0.6https://community.lizard-llm.qendryx.com/topic/19/does-any-of-my-data-leave-my-computer2026-07-19T10:34:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/6/which-graphics-cards-work-post-yours2026-07-19T10:34:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/25/two-engines-one-decision-how-lizard-routes-your-model2026-07-19T11:26:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/10/advanced-the-avx-512-vnni-option-and-why-it-s-off2026-07-20T10:17:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/22/feature-requests-and-rough-edges2026-07-20T10:17:00.000Zdaily0.6https://community.lizard-llm.qendryx.com/topic/2/lizard-llm-what-s-new-v1.2.612026-07-20T14:10:27.961Zdaily0.6https://community.lizard-llm.qendryx.com/topic/26/new-benchmark-evidence-completed-provider-comparisons2026-07-22T15:56:46.920Zdaily0.6https://community.lizard-llm.qendryx.com/topic/27/from-gguf-selection-to-local-inference-the-benchmark-setup2026-07-22T15:57:46.973Zdaily0.6https://community.lizard-llm.qendryx.com/topic/28/watch-a-native-caterpillar-and-llama.cpp-benchmark-run2026-07-22T15:58:46.994Zdaily0.6https://community.lizard-llm.qendryx.com/topic/29/what-the-latest-native-and-caterpillar-results-actually-show2026-07-22T15:59:47.010Zdaily0.6https://community.lizard-llm.qendryx.com/topic/30/technical-overview-how-lizard-native-keeps-direct3d-12-inference-resident2026-07-23T05:42:14.886Zdaily0.6https://community.lizard-llm.qendryx.com/topic/31/technical-overview-caterpillar-s-clean-room-graph-plan-execution2026-07-23T05:43:14.931Zdaily0.6https://community.lizard-llm.qendryx.com/topic/32/technical-overview-lizard-native-and-caterpillar-vs-llama.cpp-and-ollama2026-07-23T05:44:14.946Zdaily0.6https://community.lizard-llm.qendryx.com/topic/33/lizard-inference-engineering-an-inference-provider-owns-the-stack2026-07-24T01:00:10.819Zdaily0.6https://community.lizard-llm.qendryx.com/topic/34/lizard-inference-engineering-keep-model-weights-resident2026-07-24T11:00:07.552Zdaily0.6https://community.lizard-llm.qendryx.com/topic/35/lizard-chat-ask-for-a-chart-and-the-answer-arrives-with-it2026-07-24T16:32:36.228Zdaily0.6https://community.lizard-llm.qendryx.com/topic/36/remote-benchmarking-measure-the-pc-in-the-other-room-without-opening-a-port2026-07-24T16:33:36.261Zdaily0.6https://community.lizard-llm.qendryx.com/topic/37/lizard-inference-engineering-budget-against-free-vram-not-the-sticker2026-07-25T01:00:11.750Zdaily0.6https://community.lizard-llm.qendryx.com/topic/38/lizard-inference-engineering-stop-oversized-models-from-thrashing2026-07-25T11:00:06.620Zdaily0.6https://community.lizard-llm.qendryx.com/topic/39/lizard-inference-engineering-compile-shaders-once2026-07-26T01:00:06.788Zdaily0.6https://community.lizard-llm.qendryx.com/topic/40/lizard-inference-engineering-compile-the-execution-plan2026-07-26T11:00:04.260Zdaily0.6https://community.lizard-llm.qendryx.com/topic/41/lizard-inference-engineering-decode-is-a-memory-bandwidth-problem2026-07-27T01:00:11.900Zdaily0.6https://community.lizard-llm.qendryx.com/topic/42/lizard-inference-engineering-repack-q4-around-the-hot-loop2026-07-27T04:12:03.420Zdaily0.6https://community.lizard-llm.qendryx.com/topic/43/lizard-inference-engineering-more-cpu-threads-can-be-slower2026-07-28T01:00:13.749Zdaily0.6https://community.lizard-llm.qendryx.com/topic/44/lizard-inference-engineering-verify-several-drafted-tokens-in-one-pass2026-07-28T11:00:10.232Zdaily0.6https://community.lizard-llm.qendryx.com/topic/45/lizard-wants-you-100-testers-wanted2026-07-28T12:54:50.906Zdaily0.6https://community.lizard-llm.qendryx.com/topic/46/lizard-native-benchmark-cli-and-http-command-reference2026-07-28T16:30:41.378Zdaily0.6https://community.lizard-llm.qendryx.com/topic/47/caterpillar-benchmark-cli-and-http-command-reference2026-07-28T16:30:41.476Zdaily0.6https://community.lizard-llm.qendryx.com/topic/50/lizard-inference-engineering-stop-rebuilding-descriptors-per-token2026-07-29T01:00:14.434Zdaily0.6https://community.lizard-llm.qendryx.com/topic/51/lizard-inference-engineering-use-uma-without-a-fake-copy2026-07-29T11:00:12.455Zdaily0.6https://community.lizard-llm.qendryx.com/topic/52/lizard-inference-engineering-make-kv-cache-precision-visible2026-07-30T01:00:13.028Zdaily0.6https://community.lizard-llm.qendryx.com/topic/53/lizard-inference-engineering-do-not-ship-an-unproven-fast-path2026-07-30T11:00:11.470Zdaily0.6https://community.lizard-llm.qendryx.com/topic/54/lizard-inference-engineering-fail-preflight-instead-of-inventing-a-row2026-07-31T01:00:12.643Zdaily0.6https://community.lizard-llm.qendryx.com/topic/55/lizard-inference-engineering-prove-a-process-really-exited2026-07-31T11:00:11.881Zdaily0.6https://community.lizard-llm.qendryx.com/topic/56/lizard-inference-engineering-keep-oversized-remote-tests-honestly-directx2026-08-01T01:00:13.107Zdaily0.6https://community.lizard-llm.qendryx.com/topic/57/lizard-inference-engineering-missing-is-not-zero2026-08-01T11:00:10.903Zdaily0.6https://community.lizard-llm.qendryx.com/topic/58/lizard-inference-engineering-keep-the-api-local-and-familiar2026-08-02T01:00:11.338Zdaily0.6https://community.lizard-llm.qendryx.com/topic/59/lizard-inference-engineering-choose-the-next-benchmark-with-us2026-08-02T11:00:06.309Zdaily0.6https://community.lizard-llm.qendryx.com/topic/60/lizard-inference-engineering-measure-the-path-users-actually-call2026-08-03T01:00:12.593Zdaily0.6https://community.lizard-llm.qendryx.com/topic/61/lizard-inference-engineering-b16-is-the-throughput-lane-not-a-single-user-promise2026-08-03T11:00:08.853Zdaily0.6https://community.lizard-llm.qendryx.com/topic/62/lizard-inference-engineering-same-model-name-is-not-proof-of-same-weights2026-08-04T01:00:12.066Zdaily0.6https://community.lizard-llm.qendryx.com/topic/63/lizard-inference-engineering-exact-output-belongs-beside-speed2026-08-04T11:00:07.257Zdaily0.6https://community.lizard-llm.qendryx.com/topic/64/lizard-inference-engineering-separate-user-visible-speed-from-native-decode-speed2026-08-05T00:39:09.961Zdaily0.6https://community.lizard-llm.qendryx.com/topic/65/lizard-inference-engineering-ttft-and-completion-time-diagnose-different-pain2026-08-05T10:27:08.242Zdaily0.6https://community.lizard-llm.qendryx.com/topic/66/lizard-inference-engineering-let-measured-b16-and-b8-results-choose-the-lane2026-08-06T01:00:12.117Zdaily0.6https://community.lizard-llm.qendryx.com/topic/67/lizard-inference-engineering-match-performance-evidence-to-model-quant-and-context2026-08-06T01:57:03.991Zdaily0.6https://community.lizard-llm.qendryx.com/topic/68/lizard-inference-engineering-exact-token-ids-matter-at-the-prompt-boundary2026-08-06T03:13:06.046Zdaily0.6https://community.lizard-llm.qendryx.com/topic/69/lizard-inference-engineering-stable-evidence-should-survive-a-noisy-rerun2026-08-06T11:00:10.476Zdaily0.6https://community.lizard-llm.qendryx.com/topic/70/lizard-inference-engineering-capacity-still-matters-when-no-benchmark-exists2026-08-07T11:00:08.174Zdaily0.6https://community.lizard-llm.qendryx.com/topic/72/lizard-inference-engineering-unsupported-models-should-fail-before-inference2026-08-08T01:00:10.252Zdaily0.6https://community.lizard-llm.qendryx.com/topic/73/lizard-inference-engineering-critical-failures-deserve-a-ticket-warnings-do-not2026-08-09T01:00:08.477Zdaily0.6https://community.lizard-llm.qendryx.com/topic/74/lizard-inference-engineering-one-ticket-id-should-connect-the-whole-incident2026-08-09T11:00:07.435Zdaily0.6https://community.lizard-llm.qendryx.com/topic/75/lizard-inference-engineering-two-native-providers-should-remain-explicit2026-08-10T01:00:12.315Zdaily0.6https://community.lizard-llm.qendryx.com/topic/76/lizard-inference-engineering-caterpillar-compiles-the-graph-around-the-hot-loop2026-08-10T11:00:31.817Zdaily0.6https://community.lizard-llm.qendryx.com/topic/77/lizard-inference-engineering-integrated-gpus-need-a-shared-memory-plan2026-08-11T01:00:07.780Zdaily0.6https://community.lizard-llm.qendryx.com/topic/78/lizard-inference-engineering-cpu-only-should-remain-a-first-class-path2026-08-11T11:00:09.364Zdaily0.6https://community.lizard-llm.qendryx.com/topic/79/lizard-inference-engineering-use-a-familiar-local-http-contract2026-08-12T01:00:09.952Zdaily0.6https://community.lizard-llm.qendryx.com/topic/80/lizard-inference-engineering-keep-local-serving-observable2026-08-12T11:00:09.177Zdaily0.6https://community.lizard-llm.qendryx.com/topic/81/lizard-inference-engineering-warm-the-selected-model-before-the-first-conversation2026-08-13T01:00:09.212Zdaily0.6https://community.lizard-llm.qendryx.com/topic/82/lizard-inference-engineering-prompt-caching-should-follow-conversation-structure2026-08-13T11:00:10.625Zdaily0.6https://community.lizard-llm.qendryx.com/topic/83/lizard-inference-engineering-direct-benchmarks-and-http-benchmarks-answer-different-questions2026-08-14T01:00:07.734Zdaily0.6https://community.lizard-llm.qendryx.com/topic/84/lizard-inference-engineering-batch-means-concurrent-requests-not-a-compression-sweep2026-08-14T11:00:09.731Zdaily0.6https://community.lizard-llm.qendryx.com/topic/85/lizard-inference-engineering-quantization-changes-more-than-file-size2026-08-15T01:00:08.085Zdaily0.6https://community.lizard-llm.qendryx.com/topic/86/lizard-inference-engineering-unknown-quantization-should-stay-unknown2026-08-15T11:00:06.627Zdaily0.6https://community.lizard-llm.qendryx.com/topic/87/lizard-inference-engineering-model-release-must-be-proven-not-assumed2026-08-16T01:00:11.514Zdaily0.6https://community.lizard-llm.qendryx.com/topic/88/lizard-inference-engineering-a-restart-should-remember-the-last-valid-runtime2026-08-16T11:00:08.259Zdaily0.6https://community.lizard-llm.qendryx.com/topic/89/lizard-inference-engineering-onboarding-should-recommend-only-compatible-models2026-08-17T01:00:11.427Zdaily0.6https://community.lizard-llm.qendryx.com/topic/90/lizard-inference-engineering-activation-should-preserve-the-selected-provider2026-08-17T11:00:09.575Zdaily0.6https://community.lizard-llm.qendryx.com/topic/91/lizard-inference-engineering-hardware-detection-should-describe-the-configuration2026-08-18T01:00:13.452Zdaily0.6https://community.lizard-llm.qendryx.com/topic/92/lizard-inference-engineering-resource-telemetry-explains-unstable-throughput2026-08-18T11:00:11.531Zdaily0.6https://community.lizard-llm.qendryx.com/topic/93/lizard-inference-engineering-chat-metrics-should-be-scoped-to-the-active-provider2026-08-19T01:00:11.516Zdaily0.6https://community.lizard-llm.qendryx.com/topic/94/lizard-inference-engineering-context-length-belongs-in-lived-performance-history2026-08-19T11:00:13.237Zdaily0.6https://community.lizard-llm.qendryx.com/topic/95/lizard-inference-engineering-keep-user-metrics-attached-to-the-response2026-08-20T01:00:08.344Zdaily0.6https://community.lizard-llm.qendryx.com/topic/96/lizard-inference-engineering-do-not-turn-unavailable-telemetry-into-zero2026-08-20T11:00:08.394Zdaily0.6https://community.lizard-llm.qendryx.com/topic/97/lizard-inference-engineering-make-benchmark-commands-copyable-and-complete2026-08-21T01:00:10.452Zdaily0.6https://community.lizard-llm.qendryx.com/topic/98/lizard-inference-engineering-community-questions-should-keep-users-anonymous2026-08-21T11:00:11.826Zdaily0.6https://community.lizard-llm.qendryx.com/topic/99/lizard-inference-engineering-performance-leadership-must-be-re-earned-on-each-machine2026-08-22T01:00:14.264Zdaily0.6https://community.lizard-llm.qendryx.com/topic/100/lizard-inference-engineering-run-the-real-http-readiness-check2026-08-22T11:00:14.988Zdaily0.6