Cost

What it costs to run the model. Not a capability, but it decides model choice as often as any of them, and it is the axis where the ranking inverts: lower is better.

Ordered by how many of these benchmarks place a model in their top 10, then by its best placing. Each column keeps its own order; nothing here is averaged.

Model	In top 10	Artificial Analysis LLM Leaderboard Blended Price, USD per 1M tokens (3:1 input:output)	Artificial Analysis Coding Agents Cost
Apriel-v1.6-15B-Thinker servicenow	1/2	#1 $0.00	—
GPT-5.6 Luna openai	1/2	#62 $0.45	#1 $0.04
Command A+ cohere	1/2	#2 $0.00	—
DeepSeek V4 Flash deepseek	1/2	#29 $0.12	#2 $0.07
Devstral 2 mistral	1/2	#3 $0.00	—
DeepSeek V4 Pro deepseek	1/2	#68 $0.54	#3 $0.27
Devstral Small 2 mistral	1/2	#4 $0.00	—
GPT-5.6 Terra openai	1/2	#117 $4.50	#4 $0.30
G9v3-39A5B ai9star	1/2	#5 $0.00	—
Kimi K2.6 kimi	1/2	—	#5 $1.19
G9v3-3B ai9star	1/2	#6 $0.00	—
Gemini 3.7 Flash google	1/2	#96 $1.50	#6 $1.26
Gemma 4 31B google	1/2	#7 $0.00	—
Claude Opus 4.6 anthropic	1/2	—	#7 $1.28
LFM2.5-8B-A1B liquidai	1/2	#8 $0.00	—
GPT-5.6 Sol openai	1/2	#125 $11.25	#8 $1.40
LFM2.5-VL-1.6B liquidai	1/2	#9 $0.00	—
Muse Spark 1.1 meta	1/2	—	#9 $1.43
Ling 3.0 Tiny inclusionai	1/2	#10 $0.00	—
GPT-5.4 openai	1/2	—	#10 $1.55

A dash means the source does not cover that model, not that it scored zero. Models with no registry entry are excluded here because they cannot be joined across sources — they are still shown, flagged, on their own source's board.