Charts

Scores you can see

Each chart uses a number we stored. A model with no score is left off.

UsageToken volume, apps, provider speed TodayTokens by model TrendSeven-day token change WeekSeven-day token total MonthThirty-day token total ChangeDay-to-day token change DaysStored ranking days FilesStored rankings file days AuthorsTokens by author ToolsTool-calling tokens ImagesImage input tokens PicturesImage output tokens AudioAudio tokens BroadcastStored broadcast calls PromptStored broadcast prompt tokens AnswerStored broadcast completion tokens SpendStored broadcast cost SlugStored broadcast slug AppsPopular apps by tokens RequestsPopular apps by requests RisingTrending apps CallsTrending apps by requests SessionsMedian session cost SlugStored session app slug TasksTask token share UsageTask usage share GroupsTask group token share Group useTask group usage share InsideTask share of its group Inside useTask usage share of its group ModelModel token share of a task Model useModel usage share of a task LeadStored task lead share LatencyProvider latency SpeedProvider tokens per second InputProvider input price ListedCatalog prompt price FloorLong prompt price TiersStored price override count ClockTime-window prompt price Clock outTime-window output price Clock readTime-window cache read price AboveLong output price ReadLong cache read price WriteLong cache write price Long hourLong 1-hour cache write price OutputProvider output price FinishCatalog output price ReasonReasoning token price ReasonProvider reasoning price FeeWeb search price SearchProvider web search price AudioAudio token price Long audioLong audio token price VoiceAudio output price AudioProvider audio price VoiceProvider audio output price CacheAudio cache price Long listenLong audio cache price ImageImage input price ImageProvider image price PictureImage output price PictureProvider image output price CatalogCatalog cache read price DiscountCache discount percent TTLCache time to live CacheProvider cache read price WriteProvider cache write price 1 hourProvider 1-hour cache write price HourCatalog 1-hour cache write price UptimeProvider 1-day uptime 5 minProvider 5-minute uptime 30 minProvider 30-minute uptime PricePrice vs intelligence SpreadIntelligence bands New12 newest scored models FillHow full each slice is DirectionStored slice direction Slice groupStored benchmark slice group FieldStored benchmark slice field SourceStored benchmark slice source APIStored benchmarks API source SchemaStored slice schema SchemaStored benchmarks schema PortalStored benchmarks portal CountStored benchmarks count PortalStored slice portal As ofStored slice snapshot time CiteStored slice citation SlicesStored slice count RankedStored slice ranked count MissingStored slice missing count TruncatedStored slice truncated flag WindowWho has which score SlicesStored slices per model CapStored slice window cap FromStored slice window start ThroughStored slice window end ModelsStored slice window model count AnyStored slice window scored models NoneStored slice window unscored models RuleStored slice window rule MakersModels per maker CensusStored catalog counts ListedStored OpenRouter model count BoardStored leaderboard model count PlaceStored leaderboard rank UpdatedStored leaderboard time PopularStored popular model count SchemaStored popular telemetry schema UpdatedStored popular telemetry time WindowStored popular context window MaxStored popular max output PromptStored popular prompt price CompleteStored popular completion price CachedStored popular cache read price WriteStored popular cache write price ListsStored view list sizes FamilyModels per stored family ConfidenceStored confidence label StateStored catalog status RunStored run status counts PinsStored distinct pin count CodersStored coder model count JudgesStored judge model count HeadcountStored ballot judge count BallotsStored ballot summary total SpendStored ballot summary cost CrossStored cross status counts CrewStored judge crew counts RosterStored roster status counts SizeStored run size counts CloseStored ready to close DateStored hero run date SchemaStored run status schema PhaseStored run status phase DoneStored run status done flag AtStored run status time BuiltStored trials file time VersionStored trials file version OpenStored roster opencode flag BandStored band fraction limits Roster capStored roster cap list AgentsStored agent duration BlocksStored run-block duration BeganStored run-block start EndedStored run-block end TrialsStored agent trial count MinutesStored run minutes StartStored agent start time EndStored agent end time FreeStored free flag DynoEmpirical dyno score CautiousStored dyno cautious fill HighStored dyno high fill AxesStored dyno axes scored ScoreStored leaderboard empirical score HundredStored empirical 100 RatedStored rating date CoverStored coverage status BaseStored base score AwardedStored awarded score CautiousStored rating cautious fill HighStored rating high fill FactorStored coverage factor EvidenceStored evidence path ScarStored scar label DetailStored scar description ActionStored scar action Run scarsStored run-log scar counts FileStored measured scar file SampleStored measured scar judge CoderStored measured scar coder TextStored measured scar text PartnerStored wingman partner OrthStored wingman orthogonality RescueStored wingman rescue rate Rescue hiStored rescue of high given low Rescue loStored rescue of low given high CofailStored cofail rate DisagreeStored disagreement frequency BallotStored judge ballot score WhoStored who erred SplitStored ballot disagree flag NoteStored ballot note PromptStored ballot prompt tokens AnswerStored ballot completion tokens CostStored ballot dollar cost SideStored ballot seat A Side BStored ballot seat B FlagStored ballot ok flag IdStored ballot work-order id SlugStored ballot slug JudgeStored ballot judge name ExclusiveStored exclusive error rate AttributedStored attributed error rate SeenStored appearance count Seat AStored seat A count Seat BStored seat B count ErrorsStored attributed error count SignedStored signed complement CountStored disagreement count CosineStored error-vector cosine ExcessStored excess cofailure ErrorStored mean error magnitude VectorStored error-vector magnitude GeometricStored geometric orthogonality Pair orthStored pair orthogonality LowStored pair low member HighStored pair high member AxisStored pair axis TrialsStored pair trial count PairsStored pair count NamedStored telemetry host names HostsStored host count per model HostStored external page host IndexedStored external page count SectionStored external page section PathStored page path prefix PairStored two-segment page path NamedStored page model id SlugStored hero seat slug PinStored hero seat pin DossierStored hero seat dossier TierStored hero seat tier HarnessStored hero seat harness RoleStored hero seat role SeatedStored seated flag FibStored fib ok flag Host okStored hero seat host ok FullStored hero-full flag HTTPStored seat HTTP status Seat scarStored hero seat scar AliasStored OpenRouter alias target CanonicalStored canonical slug HubStored Hugging Face id AxesStored axes scored CautiousStored cautious projection HighStored high projection SchemaStored forward schema PortalStored forward portal CountStored forward projection count MakerStored projection maker CurrentStored projection current score GapStored projection gap High fillStored projection high fill SpeedGroq tokens per second GroqContext, price, file size LimitsGroq contact-sales limits ContextContext window CutoffKnowledge cutoff date EndsCatalog expiration date CreatedCatalog created date UpdatedCatalog updated time SchemaStored catalog schema PortalStored catalog portal LegacyStored legacy portal DossiersStored catalog dossier count ReleasedStored release month SizeParameter count ClassStored weight class LicenseCatalog license ArchStored architecture text SummaryStored catalog summary BlurbStored OpenRouter description LinkStored OpenRouter endpoints link PurposeStored catalog purpose ModalityCatalog modality InputsInput modalities OutputsOutput modalities TokenizerCatalog tokenizer InstructInstruct type ParamsSupported parameters CountSupported parameter count EffortReasoning effort options DefaultDefault reasoning effort RequiredMandatory reasoning FlagReasoning max-tokens flag OnReasoning on by default TempDefault temperature Top pDefault top p Top kDefault top k RepeatDefault repetition penalty FrequencyDefault frequency penalty HostsEndpoint rows by provider CatalogStored catalog endpoint rows FilledStored catalog score fill ModelsEndpoint rows by model StatusProvider endpoint status LineStored endpoint online flag ModelsStored endpoint model count RouteStored endpoint API source ModeratedTop provider moderated ProvidersProvider context window MaxProvider max completion Top maxTop provider max completion CeilingCatalog max output tokens PromptProvider max prompt tokens QuantProvider quantization TierProvider service tier TagProvider endpoint tag PairCompare two models Top 3Slice finder NamesWhat each slice means DownloadsHugging Face top 40 GoogleGoogle catalog downloads LikesGoogle catalog likes TasksGoogle catalog tasks CountGoogle catalog model count TasksHugging Face downloads by task AuthorsHugging Face downloads by author LikesHugging Face likes Task likesHugging Face likes by task Author likesHugging Face likes by author UpdatedHugging Face recently updated GatedHugging Face gated access PrivateHugging Face private flag PapersStored paper count SpecificStored model-specific paper count CategoryStored paper category TagsStored paper category tags NamedStored paper model family TargetsStored paper target families CitedStored paper benchmark name ModelsStored paper model name DatedStored paper published date OrderStored paper list index UpdatedStored paper updated date PulledStored paper extract time AuthorsStored paper author CommentStored paper comment FlagStored benchmark-paper flag SpecificStored model-specific flag DOIStored paper DOI PDFStored paper PDF link arXivStored paper arXiv id URLStored paper abstract URL QueryStored paper search query IdStored paper clean id TitleStored paper title AbstractStored paper summary InstructionsIFEval scores ScienceGPQA Diamond accuracy PlaceGPQA Diamond rank MathGSM8K accuracy MathMATH-500 accuracy CostGPQA Diamond cost per task PlaceGPQA Diamond cost rank IndexArtificial Analysis intelligence index PlaceIntelligence index rank CodingArtificial Analysis coding index PlaceCoding index rank LiveLiveCodeBench pass@1 HardLiveCodeBench hard accuracy MediumLiveCodeBench medium accuracy EasyLiveCodeBench easy accuracy QuestionsLiveCodeBench questions evaluated CodeHumanEval pass@1 TwelveStored twelve CS score TestsStored COSMOS test count PassStored coder pass rate FullStored full-seat passes FailStored full-seat fails TrialsStored coder trial count HostStored coder host-ok count FibStored coder fib-ok count FencedStored coder fenced count PrefillStored coder prefill count KeepStored coder KEEP grade DropStored coder DROP grade HoldStored coder HOLD grade 200Stored coder HTTP 200 count 429Stored coder HTTP 429 count 503Stored coder HTTP 503 count 0Stored coder HTTP 0 count 400Stored coder HTTP 400 count 403Stored coder HTTP 403 count 404Stored coder HTTP 404 count ConfirmedStored judge confirmed grades KEEPStored judge KEEP grades DROPStored judge DROP grades HOLDStored judge HOLD grades EvalsStored judge evaluation count OkStored judge ok count AverageStored judge average score OverallStored overall ballot score ScoresStored judge score counts WhoStored judge who-erred counts MaximumStored judge maximum score MinimumStored judge minimum score JudgesStored judge disagree rate SplitsStored judge disagree count BallotsStored judge ballot count SpendStored judge total cost PromptStored judge prompt tokens AnswerStored judge completion tokens LoopsStored transfer loop count ModelsStored transfer model count SeatStored transfer seat PassedStored transfer passed count FailedStored transfer failed count RateStored transfer pass rate LengthStored average output length Passed lengthStored average passed output length HTTPStored transfer HTTP counts By modelStored transfer HTTP by model ReasonStored transfer failure reason WorkStored transfer work-order loops OrderStored transfer work-order id TaskStored transfer task description OkStored transfer ok flag StampStored transfer timestamp TextStored transfer text length CodeStored loop HTTP code CodeMBPP pass@1 SWEStored SWE-bench score SWESWE-bench Verified LiteSWE-bench Lite TauStored SWE-bench tau AgenticArtificial Analysis agentic index PlaceAgentic index rank AirlineVerified airline bench accuracy PlaceVerified airline bench rank FareAirline bench cost per task PlaceAirline bench cost rank SearchBrowseComp accuracy PlaceBrowseComp rank RunSearch bench reasoning effort MetricSearch bench primary metric SurfaceSearch bench surface EngineSearch bench engine TurnsSearch bench max agent turns ValueBrowseComp cost per task PlaceBrowseComp cost rank WaitBrowseComp latency PlaceBrowseComp latency rank WideWideSearch item F1 PlaceWideSearch rank ValueWideSearch cost per task PlaceWideSearch cost rank WaitWideSearch latency PlaceWideSearch latency rank DeepDeepSearchQA accuracy PlaceDeepSearchQA rank ValueDeepSearchQA cost per task PlaceDeepSearchQA cost rank WaitDeepSearchQA latency PlaceDeepSearchQA latency rank ExamHLE search accuracy PlaceHLE search rank ValueHLE search cost per task PlaceHLE search cost rank WaitHLE search latency PlaceHLE search latency rank SpeedTokens per second BoardStored leaderboard speed FirstTime to first token ArenaHard prompts, coding, text EloStored LMSYS Arena Elo ToolsTool-calling ranks CardStored popular BFCL score BenchStored tool-calling benchmark LiveBFCL live accuracy Live simpleBFCL live simple accuracy Live multipleBFCL live multiple accuracy Live parallelBFCL live parallel accuracy Live severalBFCL live parallel multiple accuracy Non-liveBFCL non-live accuracy MultipleBFCL multiple accuracy ParallelBFCL parallel accuracy SimpleBFCL simple accuracy SeveralBFCL parallel multiple accuracy TurnsBFCL multi-turn accuracy BaseBFCL multi-turn base MissingBFCL missing function accuracy ParamsBFCL missing parameter accuracy RelevanceBFCL relevance detection IrrelevanceBFCL irrelevance detection CostBFCL total cost MeanBFCL mean latency LongBFCL long context accuracy WebBFCL web search accuracy Web baseBFCL web search base accuracy No snippetBFCL web search no snippet accuracy P95BFCL 95th percentile latency DeviationBFCL latency standard deviation MemoryBFCL memory accuracy KeysBFCL key-value memory accuracy VectorBFCL vector memory accuracy SummaryBFCL recursive memory accuracy LicenseBFCL license OrgBFCL organization CodeAider polyglot top 40 CostAider polyglot cost FirstAider first-try pass rate RateAider stored pass rate 2 FormatAider correct edit format WholeAider whole-file pass rate DiffAider diff-format pass rate CardStored popular Aider diff pass PromptAider prompt tokens AnswerAider completion tokens LabelAider edit format CommandAider stored command RunAider stored run directory DatedAider stored run date CasesAider stored test cases BoardAider stored benchmark source CatalogStored design arena Elo DesignDesign Arena top 8s BoardStored Design Arena arena CategoryStored Design Arena category WebsiteDesign Arena website Elo WinsDesign Arena website win rate PlaceDesign Arena website rank PlaceDesign Arena website speed rank 3DDesign Arena 3D Elo WinsDesign Arena 3D win rate PlaceDesign Arena 3D rank PlaceDesign Arena 3D speed rank CodeDesign Arena code Elo WinsDesign Arena code win rate PlaceDesign Arena code rank PlaceDesign Arena code speed rank GameDesign Arena game Elo WinsDesign Arena game win rate PlaceDesign Arena game rank PlaceDesign Arena game speed rank UIDesign Arena UI Elo WinsDesign Arena UI win rate PlaceDesign Arena UI rank PlaceDesign Arena UI speed rank VizDesign Arena data viz Elo WinsDesign Arena data viz win rate PlaceDesign Arena data viz rank PlaceDesign Arena data viz speed rank SVGDesign Arena SVG Elo GraphicDesign Arena graphic design Elo WinsDesign Arena graphic design win rate PlaceDesign Arena graphic design rank LogoDesign Arena logo Elo WinsDesign Arena logo win rate PlaceDesign Arena logo rank ImageDesign Arena image Elo WinsDesign Arena image win rate PlaceDesign Arena image rank EditDesign Arena image editing Elo WinsDesign Arena image editing win rate PlaceDesign Arena image editing rank WinsDesign Arena SVG win rate PlaceDesign Arena SVG rank PlaceDesign Arena SVG speed rank ASCIIDesign Arena ASCII Elo WinsDesign Arena ASCII win rate PlaceDesign Arena ASCII rank PlaceDesign Arena ASCII speed rank TimeDesign Arena generation time AgentsDesign Arena agent Elo WinsDesign Arena agent win rate PlaceDesign Arena agent rank OtherStored other ranking tokens NamedStored named ranking tokens RowsStored ranking rows per day LeadStored daily ranking leader tokens File rowsStored rankings file row counts Input rowsAudio and image ranking rows PullStored broadcast pull counts FetchedStored broadcast fetch time JoinsStored broadcast joins FilledStored endpoint coverage counts SqliteStored sqlite latency and throughput fill TablesStored sqlite table count RowsStored sqlite catalog rows TurnsStored session turn-range counts App idStored popular app id TagsStored task tag count StartStored rankings file start date EndStored rankings file end date As ofStored rankings window time VersionStored rankings window version OpensStored rankings window start date ClosesStored rankings window end date CountStored latest named model count FileStored rankings file name LicenseStored rankings dataset license SourceStored rankings dataset request CiteStored rankings dataset citation SourcesStored digest citation lines SchemaStored digest schema RuleStored digest rule GapsStored digest gap lines FilterStored rankings modality filter RowsStored app board row counts PagedStored board page count PulledStored app and session as-of time APIStored app and session version FromStored app board start date ThroughStored app board end date EndsStored session window end date LengthStored session window days RowsStored session cost rows