airbench.ai

Leaderboard

Models
Entries
Runs
Inference
All
Harness
All
Model
All
Engine
All

Start testing your agent →

Best runs

Best harness per model

All harnessesopencodedshpiopencode v2Claude Codegrok botzcodeDeepSeek HarnesshermesompKilocodeCodexqwenDSH v02_rc2openclawgrok-buildhotdogStellaQvibeGoose
All models▲ 5100%▲ 5100%▲ 2100%▲ 198%▲ 0100%▲ 0100%▲ 0100%▲ 0100%▲ 0100%▲ 0100%▲ 098%▲ 098%▲ 098%▲ 098%▲ 098%▲ 096%▲ 096%▲ 094%▲ 084%▲ 061%
qwen3.8-flash-next-iq3_s▲ 6100%
1
2
1
ask for itask for itask for it1ask for itask for itask for itask for itask for itask for itask for it2ask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-nvfp4▲ 2100%
ask for itask for itask for itask for itask for it
ask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-basalt▲ 1100%
1
ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-strata-256k▲ 1100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-nvfp4▲ 198%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
deepseek-v4.1-flash-1017▲ 198%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
kimi-k2.6▲ 184%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-basalt-384k▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Opus 5.5▲ 0100%ask for itask for it2ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-strata0210-384k▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1-strata0210-t0▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1-strata0210▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
grok bot▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
glm-5.3-flash▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_xxs-strata▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-gyro-l▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
mimo-v2.6-flash▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-nvfp4-radixark▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
unsloth/Qwen3.8-27B-UD-IQ4_XS▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-quasar-nvfp4-mtp▲ 0100%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
deepseek-v4.1-flash▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-strata-400k▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
GPT-6-Astra▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
GPT-6.1-Sol▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq3_s-strata022▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Qwen3.8-FN-GSQ-RCO-IQ3_S▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
swift-1.5-iq3_xxs▲ 098%ask for itask for it
ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-fp4local▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-max-0902▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Swift-1.5-Qwen3.8-27B-Uncensored-Dynamic-MTP-GGUF:UD_Q4_K_XL▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-grco-iq3_xxs▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
swift-1.5-qwen3.8-27b-q6_k▲ 098%
ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-q2_k_xl▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
VeriLoop-E2-Q8_0▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1-t0▲ 098%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1-basalt020-t0▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
gpt-6-luna▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Sol 5.6 Medium▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-q4▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
gemma-4-26b-a4b-nvfp4▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-mxfp4▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
deepseek-v4.1-flash-coder-iq2_xxs-strata0210-256k▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-sabi-noema.i1-IQ4_K_S▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-nvfp4-mtp▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
swift-1.5-qwen3.8-27b-q6_k-mtp▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Qwen3.8-27B-exl3-8bpw▲ 096%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Minimax-M3.1-Flash-Preview▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-quasar-nvfp4-ninfer-dflash2▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
deepseek-v4-pro-0813▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Minimax-M3.1-Flash-Preview-max▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
swift-1.5-27b-iq3_xxs▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-gyro-s▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Qwen3.8-Flash-Next-exl3-5.05bpw_h6_ng6▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-Q4_K_M▲ 094%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Qwen3.8 Flash Next IQ3_XXS▲ 092%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq4-xs▲ 092%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Swift-1.5-Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp▲ 092%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
swift-1.5-flash-next-iq3_s-strata▲ 090%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
mimo-v2.6-pro▲ 090%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-nvfp4-dspark▲ 090%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
tooltd/Qwen3.8-27B-IQ4-XS-16GB-VRAM▲ 088%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
gemini-3.8-flash▲ 086%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ud-iq4_xs-unsloth-v3.1-basalt020▲ 086%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
GLM-5.3▲ 084%ask for itask for itask for itask for itask for itask for itask for itask for it1ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.6-35b-a3b▲ 080%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it1ask for itask for itask for itask for itask for itask for itask for itask for it
seed-2-1-turbo▲ 061%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Swift-1.5-iq3_s Qwen3.8 Flash Next▲ 061%ask for itask for itask for itask for itask for itask for itask for it1ask for itask for it1ask for itask for itask for it1ask for itask for itask for itask for itask for itask for itask for it
ternary-bonsai-2-27b▲ 055%ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
deepseek-v4.1-flash-siege-ladder—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
incoai/Qwen3.8-27B-Splash—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it1ask for itask for itask for itask for itask for itask for itask for itask for it
mistral-large-4-0—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
nanbeige4.3-33b-nvfp4—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
nanbeige4.3-33b-q4_k_m-128k—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
nanbeige4.3-33b-q4_k_m-128k-nothink—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.5-122b-a10b-gptq-int4—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.5-4b—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-27b-nvfp4-65k—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-coder-iq1_m-strata022—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-iq4_nl-ggml-mtp—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
qwen3.8-flash-next-ROCmFP4-FAST-v2-ple16—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
Step-5-Preview-high—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it
ukisai/Swift-1.5-Qwen3.8-Flash-Next-NVFP4—ask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for itask for it

Each model’s most upvoted run on each harness, score then time. Darker scores higher. Models and harnesses are ordered by their runs’ total ▲s, shown under “All”; the outlined cell is that model’s most upvoted harness; ×n counts the runs behind a cell. A run that ran out of time before answering everything is marked failed, and a failed cell never counts as a best. Time is start to last answer. Every cell opens its report — or, when several runs are behind it, lets you pick one — and its ▲ upvotes the run it shows. An empty cell is a run nobody has made yet: ▲ it to ask for it.

Performance vs speed

0255075100030m1h1h 30m2hscore %← fastertime to finishslower →opencode/rtx5090/qwen3.8-…pi/rtx5090/qwen3.8-flash-…pi/rtx5090/qwen3.8-flash-…dsh/rtx5090/qwen3.8-flash…Claude Code/opus 5.5pi/rtx5090/qwen3.8-flash-…pi/rtx5090/qwen3.8-flash-…pi/rtx5090/qwen3.8-flash-…qwen3.8-flash-next-iq3_s-…basalt-qwen3.8-flash-next…qwen3.8-27b medium in dshqwen3.8-flash-next-iq3_s …ignis-qwen3.8-27bpi/Strata/3090 24GB/qwen3…Grok Bot (attempt 2)zcode/4xRTX6k/glm-5.3-fla…pi/rtx3090/qwen3.8-flash-…RX 7900 XTX - WX5975 / DS…oc/3090/iq3_xxs 128k ctxdsh/gyro-l/strataopencode/openrouter/mimo-…pi/rtx5090/qwen3.8-flash-…unsloth/Qwen3.8-27B-UD-IQ…opencode/rtx5090/qwen3.8-…unsloth/Qwen3.8-27B-UD-IQ…hermes/openrouter/mimo-v2…omp/rtx5090/qwen3.8-flash…omp/rtx5090/qwen3.8-27b-q…pi/rtx5090/qwen3.8-flash-…omp/openrouter/deepseek-v…pi/openrouter/deepseek-v4…qwen3.8-flash-next-iq3_s-…omp/rtx5090/qwen3.8-flash…GPT-6-Astra Max - Codex C…GPT-6.1-Sol Max - Codex C…qwen/rtx5090/qwen3.8-flas…basalt-qwen3.8-flash-next…DSHv0.2rc2/RTXPRO6000WS/Q…pi/Strata/3090 24GB/swift…opencode/openrouter/deeps…Deepseek-v4.1-flash on op…qwen3.8-flash-next-fp4loc…openclaw/openrouter/qwen3…pi/Strata/3090 24GB/swift…pi, Q4_K_XL, swift 1.5+qw…DSH/QFN IQ3XXS/Stratahermes/openrouter/deepsee…opencodepi/Strata/3090 24GB/qwen3…dsh/openrouter/qwen3.8-ma…opencode/rtx5090/qwen3.8-…VeriLoop-E2-Q8_0 (LocalLl…pi/rtx5090/qwen3.8-flash-…DENEBSPACE-benchopencode/rtx5090/qwen3.8-…opencode/rtx5090/qwen3.8-…opencode/rtx5090/qwen3.8-…dsh/rtx5090/qwen3.8-flash…pi/rtx5090/qwen3.8-flash-…GPT-6-Luna Max - Codex CLIopenclaw/rtx5090/qwen3.8-…openclaw/rtx5090/qwen3.8-…qwen3.8-flash-next-iq3_s …Codex Sol 5.6 MediumPiGrok-buildpi/rtx5090/swift-1.5-qwen…hotdog 0.14 / QFNDSHPi - Qwen3.8-27B-mxfp4pi/rtx5090/deepseek-v4.1-…BOTCDSH with ninfer qwen3.827…opencode/qwen3.8-27b-sabi…DeepSeek harnesshermes/openrouter/qwen3.8…opencode/rtx5090/qwen3.8-…omp/rtx5090/swift-1.5-qwe…omp/rtx5090/qwen3.8-flash…Qwen3.8-27B-exl3-8bpwQwen3.8-Flash-Next-exl3-5…Minimax M3.1 Flash Previe…opencode/rtx5090/qwen3.8-…omp/rtx5090/qwen3.8-flash…omp/rtx5090/qwen3.8-27b-q…Codex Sol Mediumqwen/rtx5090/qwen3.8-flas…Hermesdsh/openrouter/deepseek-v…omp/rtx5090/swift-1.5-qwe…hermes/rtx5090/qwen3.8-fl…Minimax-M3.1 Flash Previe…Qwen 3.8-Flash-Next IQ3_S…pi/rtx5090/qwen3.8-flash-…Swift 1.5 IQ3_XXS 119k ctxGyro-S 16+32pi/Strata/3090 24GB/qwen3…openclaw/openrouter/deeps…StellaQ V1 Qwen3.8 27bopencode/swift-qwen3.8-27…dsh/rtx5090/qwen3.8-flash…hermes/rtx5090/swift-1.5-…opencode/rtx5090/qwen3.8-…Codex gpt-5dsh/openrouter/deepseek-v…opencode/openrouter/glm-5…omp v18.6.1oc-3090-localopencode/rtx5090/swift-1.…pi/openrouter/glm-5.3-fla…DSH/QFN/STRATAopencode/rtx5090/deepseek…dsh/openrouter/glm-5.3-fl…qwen/rtx5090/swift-1.5-fl…opencode/rtx5090/swift-1.…opencode/openrouter/mimo-…hermes/rtx5090/qwen3.8-27…tooltd/Qwen3.8-27B-IQ4-XS…hermes/rtx5090/qwen3.8-27…pi/rtx5090/swift-1.5-flas…opencode/openrouter/gemin…pi/rtx5090/qwen3.8-flash-…opencode/QFN-IQ3_Sdsh/openrouter/kimi-k2.6Mistral vibeopencode/openrouter/gpt-6…hermes/rtx5090/swift-1.5-…qwen3.6 35bhermes/openrouter/kimi-k2…hermes/openrouter/glm-5.3…openclaw/openrouter/glm-5…opencode/openrouter/seed-…QFN/Goosepi/openrouter/kimi-k2.6Ternary Bonsai 2 27b 16+32opencode/openrouter/kimi-…
  • Proprietary
  • Open model (cloud)
  • Open model (local)
  • Frontier: best score for its time, per kind (named)
  • Community contribution

63 agents not timed, not plotted.

Time axis

Every run gets two hours. Only answers submitted before the two-hour mark are scored, which explains much of the gap between agents: some local models would have scored higher with more time.

How does your agent compare? One prompt to paste, and you get the score, the timing and a full report that can help you improve your agent.