Loading model…
nvidia/nemotron-3-super-120b-a12b:free7/21 in v2 — Resilience, score 72.3%, pass rate 48.4% (222/459)4.13s, output speed 6682 tok/s$0.00, prompt tokens 34573, completion tokens 15587930/30 passed (100%), score 139/15052/60 passed (86.7%), score 164/18049/60 passed (81.7%), score 235/2616/60 passed (10%), score 360/5419/60 passed (15%), score 350/54134/60 passed (56.7%), score 274/360openai/gpt-5.5 (score 77.9%, pass rate 61.9%, 284/459) and openai/gpt-5.3-codex (score 77.7%, pass rate 61.2%, 281/459); also behind moonshotai/kimi-k2.6 (score 74.8%, pass rate 60.6%, 278/459)At 7/21 in Resilience, this model is mid-tier with clear utility for cost-sensitive and speed-oriented coding tasks, but high reasoning/refactoring demands pose deployment risk. The free tier is attractive for prototyping, yet production use cases requiring reliable logic or code restructuring should consider higher-ranked alternatives.
Overall score % from merged run_models rows (chronological). Only runs that include this model appear as points.
Score % vs pass rate % per category. With 0/1 scorers, both usually line up; with proportional tests, score % reflects partial credit while pass rate counts tests that clear the fixture threshold.
Total estimated spend per scope for this model (bars, left axis) and mean spend per merged result row (line, right axis: total ÷ tests).
Pass rate % per difficulty level — complements the score % view above.
Normalized 0–100 within this model: TTFT (shorter → higher spoke) and decode tok/s (higher → higher spoke). Values come from streamed BLXBench runs merged into overall_ranking.json.
Pass rate % per category for this model (distinct from score %, which reflects partial credit).