Skip to content
Yardstick

Value

The best model is rarely the right one to buy. See what each step up in spend actually gets you, then price out your own workload.

Capability against cost

Cost per task is what a model really spent to finish one benchmark task, reasoning tokens included — a fairer yardstick than list price, because two models with the same price can think for very different lengths.

Artificial Analysis Intelligence Index against cost per taskUp and to the left is better. 28 models have both numbers.
  • On the efficient frontier
  • Beaten by a cheaper model
102030405060
Mistral Small 4
Muse Glimmer 30B
GPT-6 Luna
MiMo-V2.6-Pro
GPT-6.1 Sol
Gemini 4 Argon
GPT-6 Astra
Claude Sonnet 5.5
Claude Opus 5.5
$0.01$0.1$1$10

Cost per task, USD (log scale)

View as table
ModelCost per taskIntelligenceFrontier
Mistral Small 4$0.01511.3Yes
Muse Glimmer 30B$0.05717.5Yes
GPT-6 Luna$0.06838.1Yes
MiMo-V2.6-Pro$0.1346.3Yes
Gemini 3.5 Flash-Lite$0.1922.2—
Claude Haiku 5.5$0.2143.4—
GLM-5.3-Flash$0.2541.8—
DeepSeek-V4.1-Flash$0.2739.5—
Qwen3.8-Flash-Next$0.3739.8—
Mistral Medium 3.5$0.5014.2—
MiniMax M3$0.5129.2—
Kimi K2.7 Code$0.5425.8—
DeepSeek-V4-Pro$0.6736—
GPT-6.1 Sol$0.7251.8Yes
Qwen3.8-27B$1.0133.7—
Mistral Large 4$1.1338.4—
Gemini 3.8 Flash$1.2440.9—
Gemini 3.1 Pro$1.3029.7—
Muse Spark 1.3$1.6048.1—
Gemini 4 Argon$1.9952.6Yes
Kimi K3$2.0043.6—
GLM-5.3$2.0144.8—
GPT-6 Astra$3.2652.7Yes
Grok 4.7$3.7446.5—
Qwen3.8-Max$5.4145.4—
Claude Sonnet 5.5$5.4656Yes
Claude Opus 5.5$5.9857.6Yes
Claude Fable 5.1$7.6353.4—

The efficient frontier

For each of these, nothing cheaper scores higher on Artificial Analysis Intelligence Index. Read down the list to see what each step up in spend buys.

  1. Mistral Small 4

    Cheapest model on the frontier

    Artificial Analysis Intelligence Index
    11.3
    Cost per task
    $0.015
  2. Muse Glimmer 30B

    +6.2 Intelligence for 3.8× the cost

    Artificial Analysis Intelligence Index
    17.5
    Cost per task
    $0.057
  3. GPT-6 Luna

    +20.6 Intelligence for 1.2× the cost

    Artificial Analysis Intelligence Index
    38.1
    Cost per task
    $0.068
  4. MiMo-V2.6-Pro

    +8.2 Intelligence for 2× the cost

    Artificial Analysis Intelligence Index
    46.3
    Cost per task
    $0.13
  5. GPT-6.1 Sol

    +5.5 Intelligence for 5.4× the cost

    Artificial Analysis Intelligence Index
    51.8
    Cost per task
    $0.72
  6. Gemini 4 Argon

    +0.7 Intelligence for 2.7× the cost

    Artificial Analysis Intelligence Index
    52.6
    Cost per task
    $1.99
  7. GPT-6 Astra

    +0.1 Intelligence for 1.6× the cost

    Artificial Analysis Intelligence Index
    52.7
    Cost per task
    $3.26
  8. Claude Sonnet 5.5

    +3.3 Intelligence for 1.7× the cost

    Artificial Analysis Intelligence Index
    56
    Cost per task
    $5.46
  9. Claude Opus 5.5

    +1.6 Intelligence for 1.1× the cost

    Artificial Analysis Intelligence Index
    57.6
    Cost per task
    $5.98

Estimate your monthly bill

Enter your token volumes to rank every model by what it would cost you at list price.

Start from a workload
Input served from cache

27 models · estimated monthly cost at list price, cheapest first

  1. Claude Haiku 5.5

    Anthropic · Intelligence 43.4 · $4.00 in + $4.00 out

    $8.00per month

  2. GPT-6 Luna

    OpenAI · Intelligence 38.1 · $4.00 in + $4.00 out

    $8.00per month

  3. Qwen3.8-Flash-Next

    Alibaba · Intelligence 39.8 · $6.00 in + $3.76 out

    $9.76per month

  4. GLM-5.3-Flash

    Z.ai · Intelligence 41.8 · $6.00 in + $4.00 out

    $10.00per month

  5. Mistral Small 4

    Mistral AI · Intelligence 11.3 · $6.00 in + $4.80 out

    $10.80per month

  6. DeepSeek-V4.1-Flash

    DeepSeek · Intelligence 39.5 · $12.00 in + $9.60 out

    $21.60per month

  7. MiniMax M3

    MiniMax · Intelligence 29.2 · $12.00 in + $9.60 out

    $21.60per month

  8. MiMo-V2.6-Pro

    Xiaomi · Intelligence 46.3 · $17.40 in + $6.96 out

    $24.36per month

  9. Gemini 3.5 Flash-Lite

    Google · Intelligence 22.2 · $12.00 in + $20.00 out

    $32.00per month

  10. Qwen3.8-27B

    Alibaba · Intelligence 33.7 · $20.00 in + $24.00 out

    $44.00per month

  11. Gemini 3.8 Flash

    Google · Intelligence 40.9 · $30.00 in + $30.00 out

    $60.00per month

  12. Kimi K2.7 Code

    Moonshot AI · Intelligence 25.8 · $38.00 in + $32.00 out

    $70.00per month

  13. Muse Spark 1.3

    Meta · Intelligence 48.1 · $50.00 in + $34.00 out

    $84.00per month

  14. DeepSeek-V4-Pro

    DeepSeek · Intelligence 36 · $52.80 in + $31.68 out

    $84.48per month

  15. Mistral Large 4

    Mistral AI · Intelligence 38.4 · $54.40 in + $33.44 out

    $87.84per month

  16. GLM-5.3

    Z.ai · Intelligence 44.8 · $56.00 in + $35.20 out

    $91.20per month

  17. Mistral Medium 3.5

    Mistral AI · Intelligence 14.2 · $60.00 in + $60.00 out

    $120per month

  18. Grok 4.7

    SpaceXAI · Intelligence 46.5 · $80.00 in + $48.00 out

    $128per month

  19. Qwen3.8-Max

    Alibaba · Intelligence 45.4 · $80.00 in + $48.00 out

    $128per month

  20. Claude Sonnet 5.5

    Anthropic · Intelligence 56 · $80.00 in + $80.00 out

    $160per month

  21. GPT-6.1 Sol

    OpenAI · Intelligence 51.8 · $80.00 in + $80.00 out

    $160per month

  22. Gemini 4 Argon

    Google · Intelligence 52.6 · $80.00 in + $80.00 out

    $160per month

  23. Gemini 3.1 Pro

    Google · Intelligence 29.7 · $80.00 in + $96.00 out

    $176per month

  24. Kimi K3

    Moonshot AI · Intelligence 43.6 · $120 in + $120 out

    $240per month

  25. Claude Opus 5.5

    Anthropic · Intelligence 57.6 · $160 in + $160 out

    $320per month

  26. Claude Fable 5.1

    Anthropic · Intelligence 53.4 · $400 in + $400 out

    $800per month

  27. GPT-6 Astra

    OpenAI · Intelligence 52.7 · $400 in + $400 out

    $800per month

First-party list prices, no volume or batch discounts. Reasoning tokens are billed as output, so a model running at high effort will spend more output tokens than your visible answers suggest — count them in the output figure.

Yardstick

A plain reading of how today’s SI models compare. Scores are taken from each benchmark’s own leaderboard wherever one exists; anything a lab reported about its own model is marked.

Data as of Oct 8, 2026.

Explore

  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Methodology & sources

Sources

  • Artificial Analysis (opens in a new tab)
  • Arena (opens in a new tab)
  • Epoch AI Benchmarking Hub (opens in a new tab)
  • ARC Prize (opens in a new tab)
  • Terminal-Bench (opens in a new tab)
  • Scale AI Labs (opens in a new tab)

Yardstick is an independent reference and is not affiliated with, sponsored by or endorsed by any lab listed. Product names and logos are trademarks of their respective owners and are used only to identify them. Benchmarks are imperfect proxies: check a model on your own task before you commit to it.

  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs
  • Models
  • Compare
  • Benchmarks
  • Value
  • Labs