Skip to content
AI Atlas
BenchmarksIFBench

IFBench — cost vs performance

Best current row per canonical model in the group “accuracy · evaluator=Artificial Analysis” (334 models) against cheapest current output price (USD / 1M tokens). The dashed line is the Pareto frontier: no model is both better and cheaper than a point on it.

← Leaderboard
XOutput priceInput priceParametersContextMemory (est.)ScaleloglinearBubblecontextparamsnone

open / restricted weights closed Pareto frontier (12 models)bubble = context window

Frontier models 12

Pareto frontier
Modelaccuracy on IFBenchRankcheapest current output price (USD / 1M tokens)ProviderTrust
Grok 4.3xAI · Closed83.3%1$2xAI APIIndependent
MiniMax-M3MiniMax · Open weights82.9%3$1.2Fireworks AIIndependent
MiMo-V2.5-ProXiaomi · Open weights79.9%8$0.87OpenRouterIndependent
gpt-5.4-nanoOpenAI · Closed75.9%23$0.625OpenAI APIIndependent
Gemma 4 31BGoogle · Open weights75.6%28$0.34Google Gemini APIIndependent
Gemma 4 26B A4BGoogle · Open weights72.5%46$0.22Google Gemini APIIndependent
Nemotron 3 Nano 30B A3BNVIDIA · Open weights71.1%57$0.20OpenRouterIndependent
gpt-oss-120bOpenAI · Open weights69.0%67$0.17OpenRouterIndependent
Qwen3.5-9BQwen · Open weights66.7%81$0.15OpenRouterIndependent
gpt-oss-20bOpenAI · Open weights65.1%88$0.13OpenAI APIIndependent
Gemma 3 4BGoogle · Restricted weights28.3%287$0.10OpenRouterIndependent
Mistral Small 3Mistral AI · Open weights26.4%297$0.08Mistral AI La PlateformeIndependent

Methodology. Points are the best current row per canonical model in comparability group 'accuracy · evaluator=Artificial Analysis'. Price = cheapest current offer across providers (the provider shown). Pareto frontier maximises the score and minimises x; exact ties are all kept. memory_estimate is an estimate (see /methodology). Only the selected comparability group is plotted; points under other configurations are not mixed in. Price = cheapest current offer across providers at the time of the last crawl. Nothing is estimated except where marked.