Leaderboards
Benchmarks
Every score carries its origin — leaderboard is independent,vendor is self-reported. History is kept, never overwritten.
LiveBench Mathematics
% average over LiveBench competition and olympiad maths taskshigher is betterProvenanceindependent leaderboard35 modelsscores Jun 25, 2026
1GPT-5.6 SolOpenAI96.2via Max6—Jun 25, 2026leaderboard2Claude Fable 5Anthropic95.99via Max effort2—Jun 25, 2026leaderboard3GPT-5.5OpenAI95.86via xHigh10—Jun 25, 2026leaderboard4Claude Opus 5Anthropic95.73via Max effort4—Jun 25, 2026leaderboard5GPT-5.6 TerraOpenAI94.91via Max32—Jun 25, 2026leaderboard6Claude Opus 4.8Anthropic94.32via Max effort4—Jun 25, 2026leaderboard7GPT-5.4OpenAI94.15via xHigh13—Jun 25, 2026leaderboard8GPT-5.2 ChatOpenAI93.17via High13—Jun 25, 2026leaderboard9Claude Sonnet 5Anthropic92.94via xHigh effort9—Jun 25, 2026leaderboard10Claude Opus 4.7Anthropic92.85via xHigh effort4—Jun 25, 2026leaderboard11GPT-5.4 NanoOpenAI90.98via xHigh146—Jun 25, 2026leaderboard12Grok 4.5xAI90.8215—Jun 25, 2026leaderboard13DeepSeek V4 ProDeepSeek1.6T90.68104~0.09Jun 25, 2026leaderboard14Claude Opus 4.5Anthropic90.39via Thinking 64k High effort4—Jun 25, 2026leaderboard15GLM 5.2Z.AI753B89.7860~0.19Jun 25, 2026leaderboard16Claude Opus 4.6Anthropic89.32via Thinking Auto High effort4—Jun 25, 2026leaderboard17GPT-5.2-CodexOpenAI88.776—Jun 25, 2026leaderboard18InklingThinking Machines952B88.36via xHigh22~0.15Jun 25, 2026leaderboard19Gemini 3.5 FlashGoogle88.24via High20—Jun 25, 2026leaderboard20GPT-5.6 LunaOpenAI87.2via Max66—Jun 25, 2026leaderboard21Muse Spark 1.1Meta87.14via xHigh21—Jun 25, 2026leaderboard22Claude Sonnet 4.6Anthropic86.99via Thinking Auto Medium effort6—Jun 25, 2026leaderboard23Gemini 3.6 FlashGoogle86.4via High23—Jun 25, 2026leaderboard24Qwen3.7 MaxQwen85.2523—Jun 25, 2026leaderboard25Kimi K3Moonshot AI2.8T84.446~0.048Jun 25, 2026leaderboard26Grok 4.3xAI84.3434—Jun 25, 2026leaderboard27Kimi K2.6Moonshot AI1.1T84.28via Thinking34~0.13Jun 25, 2026leaderboard28Qwen3.6 PlusQwen83.7243—Jun 25, 2026leaderboard29Qwen3.6 27BQwen27.8B79.8742~4.6Jun 25, 2026leaderboard30DeepSeek V4 FlashDeepSeek291B79.65452~0.43Jun 25, 2026leaderboard31Kimi K2.7 CodeMoonshot AI1.1T79.623~0.12Jun 25, 2026leaderboard32GPT-5.4 MiniOpenAI78.46via xHigh35—Jun 25, 2026leaderboard33Grok Build 0.1xAI78.4339—Jun 25, 2026leaderboard34MiniMax M3MiniMax427B76.9580~0.29Jun 25, 2026leaderboard35Gemini 3.5 Flash LiteGoogle73.74via High59—Jun 25, 2026leaderboard
A tilde in Score / GB VRAM means the memory figure behind it is calculated from the parameter count rather than read from a published file. Those ratios are shown to two significant figures, and rows within a few percent of each other are not distinguishable.