AI & LLM
Models, prompts, tokens, embeddings, evaluation, safety
450 icons · 10 groupsOpen in browser
base-modelcascadecheap-modeldecoder-onlydepthdiffusiondiffusion-modeldraft-modeledge-modelencoder-decoderencoder-onlyensemblefallback-modelfrontier-modelinstruct-modellatentllmmodelmodel-addmodel-alertmodel-aliasmodel-archivemodel-bookmarkmodel-cardmodel-checkmodel-deploymodel-downloadmodel-failmodel-familymodel-fastmodel-forkmodel-guardmodel-haltmodel-heartmodel-idlemodel-keymodel-licensemodel-lockmodel-offmodel-registry-entrymodel-scopemodel-searchmodel-sizemodel-swapmodel-tagmodel-weightsmulti-headpin-modelquantize-4bitquantized-modelreasoning-modelregion-modelsmall-language-modelsparsitytiny-modeltransformertransformer-blockvision-encoderweight-pruneattentionattention-headattention-maskattention-sinkaudio-denoisebatch-inferencebeam-searchbyte-paircite-sourcescode-interpreterconfidenceconfidence-scorecontext-extendcontext-lengthcontext-scratchcontradictioncross-attentiondecoderdetokenizediff-textdraft-verifyembedding-compareembedding-driftexpert-routerface-detectfast-thinkinggaussian-splatgreedygreedy-decodeimage-captionimage-variationinferenceinference-batchinference-fastknowledge-cutoffkv-cachekv-evictlatencylip-syncllm-chatlogit-biaslogit-lenslogitslogprobsmesh-genmin-pmodel-filtermodel-latencymodel-messagemodel-runmodel-textmodel-voicemusic-genneedle-haystacknegative-promptnerfobject-detectocr-aioutline-genoutpaintparametersparaphrasepartial-answerphysics-simplaygroundpoint-cloudpoolingpose-estimatepositional-encodingprobability-curvereasoning-budgetreasoning-traceretryrobot-armsample-consensussamplingseedsegment-maskself-consistencyself-reflectsoftmaxsound-effectspeaker-diarizespecial-tokenspeculative-decodespeech-recognizespeech-synthstop-reasonstop-sequencestreaming-responsestructured-outputstyle-transfertab-completetemperaturetemperature-hightemperature-lowtext-to-videotexture-genthink-longerthroughputtoken-ceilingtoken-intoken-outtokens-usedtop-ptranscribe-livetranslate-speechtree-of-thoughttxt2imguncertaintyupscale-aivideo-frame-aivisual-groundingvocabularyvoice-clonechain-of-thoughtfew-shotinstructionpromptprompt-chainprompt-libraryprompt-stuffingprompt-templatesystem-promptuser-promptzero-shotcost-per-tokentokentoken-costtoken-counttoken-limittoken-usagetokenizerbelief-statecheckpoint-statecompress-contextcontext-budget-alertcontext-cachecontext-compresscontext-fullcontext-pincontext-prefixcontext-routercontext-windowconversation-bufferentity-memoryevent-sourcing-agentfact-storegraph-memoryhierarchical-memorymemory-auditmemory-conflictmemory-decaymemory-evictmemory-forgetmemory-indexmemory-mergememory-namespacememory-promotememory-quotamemory-retrievememory-tiernote-to-selfpin-factprefix-cacheprompt-cache-hitprompt-cache-missreplay-logreplay-steprestore-staterolling-summaryscratchpad-notesession-stateshared-memorysliding-attentionsnapshot-agentstale-memorystate-diffstate-lockstate-machine-agentsummarise-turntoken-budgettruncate-contextturn-windowvector-memoryworld-modeldimensiondistanceembedding-modelembedding-spaceencodersimilarityvectoractivationadapteradd-checkpointalignmentcalibrationcheckpointcheckpoint-alertcheckpoint-comparecheckpoint-loadcheckpoint-savedatasetdistildpodropoutearly-stopepocheval-harnessfind-checkpointfine-tunefine-tuningfrozen-checkpointfrozen-layergrad-flowgradientgradient-clipgradient-descentgrokkinglabelled-checkpointlearning-ratelora-mergelossloss-curvelr-schedulemodel-checkpointmodel-goalmodel-trendnext-milestonenoisenoise-schedulenormalizeoptimiseroverfitoverfit-gapprecision-recallpretrainprunequantizationrank-adapterregulariseremove-checkpointresume-checkpointrlhfroc-curvestale-checkpointstop-patiencetrain-losstrainingtraining-dataunderfitunderfit-gapwarmupablationaccuracybenchmarkbest-of-nconfusion-matrixdegradationdrifteval-suiteevaluationf1golden-sethuman-evalmodel-comparisonperplexityprecisionreward-modelabstain-answeradversarial-inputalignment-taxannotation-taskbenchmark-suitebiasbias-auditbias-checkcalibration-curvecanary-promptchain-verifyconfidence-thresholdconsistency-checkcontamination-checkcontent-filterdemographic-parityerror-analysisescalation-humaneval-costeval-dataseteval-reporteval-sliceeval-tracefactuality-testfailure-taxonomyfairness-metricgold-setgroundedness-scoreguardrailguardrail-inputguardrail-outputhallucinationhallucination-detecthuman-preferencehuman-review-queueinter-annotatorjailbreak-attemptjudge-modeljudge-rubriclabel-qualityleakage-checkoversight-looppolicy-violationprompt-injection-blockrefusalrefusal-raterefusal-responsereward-hackingrubric-scoresafe-completionsafety-categorysafety-checkseverity-labelslice-analysisspecification-gamingtoxicitytoxicity-scoreuncertainty-flagaspect-cropaudio-diariseaudio-embedaudio-encoderaudio-genaudio-modelaudio-timestampaudio-to-textaudio-transcribeaudio-vadbackground-removechart-understandcross-modal-searchdenoisedepth-mapdocument-layoutface-blurframe-extractframe-interpolateimage-editimage-genimage-generationimage-maskimage-segmentimage-to-textkeypointlip-sync-aimesh-generatemodality-mixmodality-routermotion-transfermultimodalmultimodal-promptmusic-generatenoise-suppressocrocr-extractphonemeposeprosodyrender-queuescene-detectsegment-anythingshot-boundaryspeaker-diarisespeechspeech-alignspeech-to-textsubtitle-burntable-extract-aitext-to-speechtexture-generatethree-d-generatethumbnail-pickupscale-imagevideo-captionvideo-genvideo-generatevideo-summarisevideo-trackvision-modelvoice-presetwake-word
Every AI & LLM icon, free to ship
450 AI & LLM icons in outline and duotone at three weights, generated from one grid so nothing in the set can drift out of step. MIT licensed — commercial use, no attribution, no seat count.