Terminal-Bench

Source-qualified evaluations, report statistics and safety observations.

openai/gpt-6-sol

MetricValueUnit
Terminal-Bench · accuracy0.4939percent
Terminal-Bench · avg_trial_duration_sec2662.2seconds
Terminal-Bench · pass_at_20.5833percent
Terminal-Bench · pass_at_30.6242percent
Terminal-Bench · pass_at_40.6485percent
Terminal-Bench · pass_at_50.6667percent
Terminal-Bench · total_cost_usd1451.32USD

qwen/qwen3.8-max-0902

MetricValueUnit
Terminal-Bench · accuracy0.2697percent
Terminal-Bench · avg_trial_duration_sec9953seconds
Terminal-Bench · pass_at_20.3924percent
Terminal-Bench · pass_at_30.4591percent
Terminal-Bench · pass_at_40.5percent
Terminal-Bench · pass_at_50.5303percent
Terminal-Bench · total_cost_usd4635.67USD

z-ai/glm-5.3-flash

MetricValueUnit
Terminal-Bench · accuracy0.3576percent
Terminal-Bench · avg_trial_duration_sec6209.8seconds
Terminal-Bench · pass_at_20.4652percent
Terminal-Bench · pass_at_30.5136percent
Terminal-Bench · pass_at_40.5424percent
Terminal-Bench · pass_at_50.5606percent
Terminal-Bench · total_cost_usd118.73752448USD

meta/muse-wonder-1.0

MetricValueUnit
Terminal-Bench · accuracy0.4697percent
Terminal-Bench · avg_trial_duration_sec2827.5seconds
Terminal-Bench · pass_at_20.5924percent
Terminal-Bench · pass_at_30.6545percent
Terminal-Bench · pass_at_40.6909percent
Terminal-Bench · pass_at_50.7121percent
Terminal-Bench · total_cost_usd2452.31USD

openai/gpt-5.4-mini

MetricValueUnit
Terminal-Bench · accuracy0.0182percent
Terminal-Bench · avg_trial_duration_sec1848.5seconds
Terminal-Bench · pass_at_20.0318percent
Terminal-Bench · pass_at_30.0424percent
Terminal-Bench · pass_at_40.0515percent
Terminal-Bench · pass_at_50.0606percent
Terminal-Bench · total_cost_usd671.57USD

anthropic/claude-haiku-5-5

MetricValueUnit
Terminal-Bench · accuracy0.4percent
Terminal-Bench · avg_trial_duration_sec4059seconds
Terminal-Bench · pass_at_20.5167percent
Terminal-Bench · pass_at_30.5727percent
Terminal-Bench · pass_at_40.603percent
Terminal-Bench · pass_at_50.6212percent
Terminal-Bench · total_cost_usd952.47USD

openrouter/thinkingmachines/inkling

MetricValueUnit
Terminal-Bench · accuracy0percent
Terminal-Bench · avg_trial_duration_sec1402.6seconds
Terminal-Bench · pass_at_20percent
Terminal-Bench · pass_at_30percent
Terminal-Bench · pass_at_40percent
Terminal-Bench · pass_at_50percent
Terminal-Bench · total_cost_usd1011.52USD

openrouter/thinkingmachines/inkling-small

MetricValueUnit
Terminal-Bench · accuracy0.0091percent
Terminal-Bench · avg_trial_duration_sec475.1seconds
Terminal-Bench · pass_at_20.0167percent
Terminal-Bench · pass_at_30.0227percent
Terminal-Bench · pass_at_40.0273percent
Terminal-Bench · pass_at_50.0303percent
Terminal-Bench · total_cost_usd49.55USD

openrouter/reflection/beam-1.0-501b-a23b

MetricValueUnit
Terminal-Bench · accuracy0.0636percent
Terminal-Bench · avg_trial_duration_sec1440.1seconds
Terminal-Bench · pass_at_20.0879percent
Terminal-Bench · pass_at_30.0985percent
Terminal-Bench · pass_at_40.103percent
Terminal-Bench · pass_at_50.1061percent
Terminal-Bench · total_cost_usd233USD

openrouter/nvidia/nemotron-3.5-super-vl

MetricValueUnit
Terminal-Bench · accuracy0.003percent
Terminal-Bench · avg_trial_duration_sec1780.6seconds
Terminal-Bench · pass_at_20.0061percent
Terminal-Bench · pass_at_30.0091percent
Terminal-Bench · pass_at_40.0121percent
Terminal-Bench · pass_at_50.0152percent
Terminal-Bench · total_cost_usd167.29USD

anthropic/claude-fable-5-2

MetricValueUnit
Terminal-Bench · accuracy0.6576000000000001percent
Terminal-Bench · avg_trial_duration_sec7333seconds
Terminal-Bench · pass_at_20.7273percent
Terminal-Bench · pass_at_30.7545percent
Terminal-Bench · pass_at_40.7727percent
Terminal-Bench · pass_at_50.7879percent
Terminal-Bench · total_cost_usd16492.76USD

meta/muse-ember-1.0

MetricValueUnit
Terminal-Bench · accuracy0.3061percent
Terminal-Bench · avg_trial_duration_sec2288.4seconds
Terminal-Bench · pass_at_20.4303percent
Terminal-Bench · pass_at_30.503percent
Terminal-Bench · pass_at_40.5545percent
Terminal-Bench · pass_at_50.5909percent
Terminal-Bench · total_cost_usd157.18USD

openrouter/deepseek/deepseek-v4-pro-0813

MetricValueUnit
Terminal-Bench · accuracy0.0667percent
Terminal-Bench · avg_trial_duration_sec3140seconds
Terminal-Bench · pass_at_20.1212percent
Terminal-Bench · pass_at_30.1652percent
Terminal-Bench · pass_at_40.2percent
Terminal-Bench · pass_at_50.2273percent
Terminal-Bench · total_cost_usd727.02USD

openrouter/minimax/minimax-m3

MetricValueUnit
Terminal-Bench · accuracy0.0727percent
Terminal-Bench · avg_trial_duration_sec3507seconds
Terminal-Bench · pass_at_20.1348percent
Terminal-Bench · pass_at_30.1864percent
Terminal-Bench · pass_at_40.2273percent
Terminal-Bench · pass_at_50.2576percent
Terminal-Bench · total_cost_usd1149.33USD

openrouter/moonshotai/kimi-k3

MetricValueUnit
Terminal-Bench · accuracy0.2242percent
Terminal-Bench · avg_trial_duration_sec3351seconds
Terminal-Bench · pass_at_20.3485percent
Terminal-Bench · pass_at_30.4273percent
Terminal-Bench · pass_at_40.4848percent
Terminal-Bench · pass_at_50.5303percent
Terminal-Bench · total_cost_usd1919.83USD

openrouter/tencent/hy4

MetricValueUnit
Terminal-Bench · accuracy0.3909percent
Terminal-Bench · avg_trial_duration_sec4881.6seconds
Terminal-Bench · pass_at_20.4955percent
Terminal-Bench · pass_at_30.55percent
Terminal-Bench · pass_at_40.5848percent
Terminal-Bench · pass_at_50.6061percent
Terminal-Bench · total_cost_usd347.6USD

openrouter/xiaomi/mimo-v2.6-flash

MetricValueUnit
Terminal-Bench · accuracy0.1606percent
Terminal-Bench · avg_trial_duration_sec5471.6seconds
Terminal-Bench · pass_at_20.2576percent
Terminal-Bench · pass_at_30.3182percent
Terminal-Bench · pass_at_40.3576percent
Terminal-Bench · pass_at_50.3788percent
Terminal-Bench · total_cost_usd194.6USD

openrouter/xiaomi/mimo-v2.6-pro

MetricValueUnit
Terminal-Bench · accuracy0.2697percent
Terminal-Bench · avg_trial_duration_sec5265.3seconds
Terminal-Bench · pass_at_20.3818percent
Terminal-Bench · pass_at_30.447percent
Terminal-Bench · pass_at_40.4939percent
Terminal-Bench · pass_at_50.5303percent
Terminal-Bench · total_cost_usd282.32USD

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace