Cost to run a full benchmark suite
What Artificial Analysis actually paid to run its whole Intelligence Index on each configuration: a fixed batch of work, billed at list price. The spread between models is far wider than their per-token prices, because verbose reasoning multiplies the token count.
As of 21 Sept 2026, BenchLeader ranks Granite 4.2 3B first on the Cost to run a full benchmark suite board at $0.0060, ahead of GPT-5.6 Luna ($0.0098) and GPT-5.6 Luna ($0.010).
142 models · data as of 21 Sept 2026
Quality against what it costs to get there
OpenAI31
Anthropic18
Alibaba15
Google12
Mistral AI8
SpaceXAI7
DeepSeek6
Meta5
Zhipu AI5
Moonshot AI4
NVIDIA4
IBM2
MiniMax2
Xiaomi2
Ant Group1
Arcee AI1- Celeris1
Meituan1
The dashed staircase is the efficiency frontier: at each price, the best index anyone reaches. A model above and left of the rest gives you more quality per dollar of real work. Size bands and run costs come from Artificial Analysis, which is the only source that publishes what a fixed batch of work actually cost.
- 1Granite 4.2 3B$0.0060
- 2GPT-5.6 Luna (low)$0.0098
- 3gpt-oss-20b (high)$0.011
- 4Nemotron 3 Nano 30B A3B (thinking)$0.017
- 5DeepSeek V3$0.020
- 6Granite 4.2 8B$0.024
- 7Mistral Small 3.1$0.033
- 8Gemini 3.1 Flash Lite$0.039
- 9Ministral 3 3B$0.042
- 10Mistral Small 4 (thinking)$0.045
- 11Celeris-1$0.050
- 12GPT-5 mini (high)$0.053
- 13MiMo-V2.5-Pro$0.054
- 14Muse Glimmer (high)$0.055
- 15LongCat 2.0$0.059
| # | ||||
|---|---|---|---|---|
| 1 | Claude Fable 5thinkingAnthropic | $8.75 | 70.8 | $20.00 |
| 2 | Qwen3 8maxAlibaba | $5.41 | 66.5 | $2.67 |
| 3 | Claude Opus 4.8maxAnthropic | $4.08 | 64.1 | $10.00 |
| 4 | Claude Sonnet 4.6maxAnthropic | $2.49 | 57.5 | $6.00 |
| 5 | Claude Fable 5.1lowAnthropic | $2.37 | 67.4 | $20.00 |
| 6 | Qwen3.8 2.4T A95BAlibabaopen ↗ | $2.16 | 65.1 | $3.00 |
| 7 |  Quasar 438BMultiverse Computing | $2.02 | 57.1 | $0.900 |
| 8 | GLM 5.3maxZhipu AIopen ↗ | $2.01 | 65.8 | $2.15 |
| 9 | Kimi K3maxMoonshot AIopen ↗ | $2.00 | 67.1 | $6.00 |
| 10 | Gemini 3.5 FlashhighGoogle | $1.56 | 63.5 | $3.38 |
| 11 | Muse Spark 1.1xhighMeta | $1.38 | 61.7 | $2.00 |
| 12 | Muse Spark 1.3xhighMeta | $1.37 | 67.9 | $2.00 |
| 13 | Qwen3 7maxAlibaba | $1.15 | 60.9 | $3.75 |
| 14 | Claude Opus 5lowAnthropic | $1.10 | 63.3 | $10.00 |
| 15 | Nemotron 3 SuperthinkingNVIDIAopen ↗ | $1.06 | 51.3 | $0.350 |
| 16 | Grok 4.5highSpaceXAI | $1.04 | 60.9 | $3.00 |
| 17 | Muse Spark 1.2xhighMeta | $0.975 | 64.0 | $2.00 |
| 18 | GLM 5.2maxZhipu AIopen ↗ | $0.965 | 63.8 | $2.15 |
| 19 | Gemini 3.8 FlashmediumGoogle | $0.931 | 64.8 | $1.50 |
| 20 | Gemini 3.6 FlashhighGoogle | $0.929 | 61.7 | $1.50 |
| 21 | Gemini 3.7 FlashhighGoogle | $0.925 | 64.5 | $1.50 |
| 22 | GLM 5.1thinkingZhipu AIopen ↗ | $0.922 | 57.7 | $2.15 |
| 23 | GPT-5.5mediumOpenAI | $0.902 | 64.5 | $11.25 |
| 24 | Qwen3.8 27BxhighAlibabaopen ↗ | $0.820 | 59.1 | $1.06 |
| 25 | GPT-6 AstralowOpenAI | $0.818 | 68.2 | $20.00 |
| 26 | Kimi K2.6Moonshot AIopen ↗ | $0.800 | 60.4 | $1.71 |
| 27 | New Step 5 PreviewStepFun | $0.716 | 64.9 | $1.43 |
| 28 | GPT-5.5 InstantOpenAI | $0.691 | 59.0 | $11.25 |
| 29 | Gemini 3.1 ProGoogle | $0.675 | 63.8 | $4.50 |
| 30 | DeepSeek V4 PromaxDeepSeekopen ↗ | $0.674 | 64.1 | $0.544 |
| 31 | Qwen3.6 27BthinkingAlibabaopen ↗ | $0.621 | 54.8 | $1.35 |
| 32 | Nemotron 3 Ultra 550B A55BthinkingNVIDIAopen ↗ | $0.579 | 57.5 | $1.00 |
| 33 | Claude Sonnet 4.5thinkingAnthropic | $0.557 | 53.3 | $6.00 |
| 34 | Qwen3 Coder NextAlibabaopen ↗ | $0.552 | 42.5 | $0.450 |
| 35 | Kimi K2.7 CodeMoonshot AIopen ↗ | $0.541 | 55.8 | $1.71 |
| 36 | Claude Sonnet 5lowAnthropic | $0.509 | – | $4.00 |
| 37 | MiniMax M3MiniMaxopen ↗ | $0.508 | 57.4 | $0.525 |
| 38 | Grok 4.6lowSpaceXAI | $0.475 | 60.9 | $3.00 |
| 39 | Qwen3.6 35B A3BthinkingAlibabaopen ↗ | $0.475 | 53.0 | $0.557 |
| 40 | Qwen3.5 397B A17BthinkingAlibabaopen ↗ | $0.475 | 56.0 | $0.387 |
| 41 | Mistral Medium 3.5Mistral AIopen | $0.437 | 51.2 | $3.00 |
| 42 | GPT-5.4 MinixhighOpenAI | $0.410 | 55.7 | $1.69 |
| 43 | Mistral Medium 3Mistral AI | $0.392 | 45.5 | $0.800 |
| 44 | Qwen3.8-Flash-NextAlibabaopen ↗ | $0.372 | 61.5 | $0.230 |
| 45 | Trinity LargethinkingArcee AIopen | $0.353 | 47.6 | $0.413 |
| 46 | Qwen3.7 PlusAlibaba | $0.325 | 59.8 | $1.13 |
| 47 | Qwen3.5-122B-A10BthinkingAlibabaopen ↗ | $0.325 | 53.4 | $1.10 |
| 48 | Deepseek v4 Flash VisionmaxDeepSeek | $0.314 | 59.7 | $0.262 |
| 49 | Ring-2.6-1TAnt Groupopen | $0.287 | 50.0 | $0.850 |
| 50 | DeepSeek V4.1 FlashmaxDeepSeekopen ↗ | $0.265 | 61.8 | $0.262 |
| 51 | DeepSeek R1DeepSeekopen ↗ | $0.262 | 48.7 | $0.315 |
| 52 | GPT-5.6 SollowOpenAI | $0.261 | 63.6 | $8.00 |
| 53 | GLM 5.3 FlashZhipu AIopen ↗ | $0.253 | 63.9 | $0.238 |
| 54 | Gemini 2.5 ProGoogle | $0.232 | 54.4 | $3.44 |
| 55 | DeepSeek V4 FlashmaxDeepSeekopen ↗ | $0.220 | 62.0 | $0.262 |
| 56 | Claude Haiku 4.5thinkingAnthropic | $0.208 | 47.4 | $2.00 |
| 57 | GPT-5.4 NanoxhighOpenAI | $0.183 | 54.5 | $0.463 |
| 58 | GPT-5.6 Terrano reasoningOpenAI | $0.140 | 50.7 | $4.50 |
| 59 | Ministral 3 14BMistral AIopen | $0.139 | 40.0 | $0.200 |
| 60 | Grok 4.3no reasoningSpaceXAI | $0.137 | 46.6 | $1.56 |
Cite as: BenchLeader, “Cost to run a full benchmark suite”, https://www.benchleader.com/leaderboards/cost-per-run, data as of 21 Sept 2026.