modelgrep

Best OpenAI Models for Agents

Quick answer · Updated June 2026

GPT-5.5 is the best OpenAI model for agents, scoring 69.4 on the Artificial Analysis Agentic Index for tool use and multi-step task completion. GPT-5.4 (68.0) and GPT-5.3-Codex (60.5) round out the top three.

69.4Agentic
56.7Intelligence
36 t/sSpeed
$5.00Input /M
1.1MContext

AI models ranked by the Artificial Analysis Agentic Index — measuring multi-step tool use, planning and task completion (including Tau²-Bench). The best models for building autonomous agents and agentic workflows.

  1. 1O
    gpt-5.5
    ReasoningToolsJSON+156.7 intel · $5.00/M · 36 t/s
    69.4
    Agentic
  2. 2O
    gpt-5.4
    ReasoningToolsJSON+156.8 intel · $2.50/M · 40 t/s
    68.0
    Agentic
  3. 3O
    gpt-5.3-codex
    ReasoningToolsJSON+153.6 intel · $1.75/M · 50 t/s
    60.5
    Agentic
  4. 4O
    gpt-5.2-codex
    ReasoningToolsJSON+149.0 intel · $1.75/M · 49 t/s
    56.5
    Agentic
  5. 5O
    gpt-5.2
    ReasoningToolsJSON+146.6 intel · $1.75/M · 42 t/s
    54.9
    Agentic
  6. 6O
    gpt-5-codex
    ReasoningToolsJSON+144.6 intel · $1.25/M · 61 t/s
    52.7
    Agentic
  7. 7O
    gpt-5.1
    ReasoningToolsJSON+147.7 intel · $1.25/M · 50 t/s
    51.3
    Agentic
  8. 8O
    gpt-5.1-codex
    ReasoningToolsJSON+143.1 intel · $1.25/M · 39 t/s
    50.7
    Agentic
  9. 9O
    gpt-5.4-nano
    ReasoningToolsJSON+144.0 intel · $0.200/M · 63 t/s
    47.6
    Agentic
  10. 10O
    gpt-5
    ReasoningToolsJSON+142.0 intel · $1.25/M · 56 t/s
    45.8
    Agentic
  11. 11O
    gpt-5-mini
    ReasoningToolsJSON+138.9 intel · $0.250/M · 66 t/s
    40.9
    Agentic
  12. 12O
    gpt-5.1-codex-mini
    ReasoningToolsJSON+138.6 intel · $0.250/M · 135 t/s
    38.7
    Agentic
  13. 13O
    gpt-oss-120b:free
    ReasoningTools33.3 intel · Free/M · 281 t/s
    37.9
    Agentic
  14. 14O
    gpt-oss-120b
    ReasoningToolsJSON33.3 intel · $0.039/M · 281 t/s
    37.9
    Agentic
  15. 15O
    o3
    ReasoningToolsJSON+138.4 intel · $2.00/M · 67 t/s
    36.1
    Agentic
  16. 16O
    o4-mini
    ReasoningToolsJSON+133.1 intel · $1.10/M · 67 t/s
    36.1
    Agentic
  17. 17O
    gpt-oss-20b:free
    ReasoningTools24.5 intel · Free/M · 131K ctx
    27.6
    Agentic
  18. 18O
    gpt-oss-20b
    ReasoningToolsJSON24.5 intel · $0.029/M · 344 t/s
    27.6
    Agentic
  19. 19O
    gpt-4.1
    ToolsJSONVision26.3 intel · $2.00/M · 48 t/s
    27.3
    Agentic
  20. 20O
    gpt-4.1-mini
    ToolsJSONVision22.9 intel · $0.400/M · 47 t/s
    25.2
    Agentic
  21. 21O
    gpt-5.4-mini
    ReasoningToolsJSON+123.3 intel · $0.750/M · 72 t/s
    25.0
    Agentic
  22. 22O
    o3-mini-high
    ReasoningToolsJSON25.2 intel · $1.10/M · 200K ctx
    20.9
    Agentic
  23. 23O
    gpt-5-nano
    ReasoningToolsJSON+125.9 intel · $0.050/M · 91 t/s
    16.8
    Agentic
  24. 24O
    gpt-4o-2024-08-06
    ToolsJSONVision18.6 intel · $2.50/M · 23 t/s
    9.7
    Agentic
  25. 25O
    gpt-4o-2024-11-20
    ToolsJSONVision17.3 intel · $2.50/M · 64 t/s
    8.4
    Agentic

Frequently asked

What is the best OpenAI model for agents?

GPT-5.5 is the best OpenAI model for agents, scoring 69.4 on the Artificial Analysis Agentic Index for tool use and multi-step task completion. GPT-5.4 (68.0) and GPT-5.3-Codex (60.5) round out the top three.

What's a good alternative to GPT-5.5?

GPT-5.4 (68.0) is the closest alternative on this metric, followed by GPT-5.3-Codex (60.5). See the full ranking above for the tradeoffs.

How many OpenAI models are there?

modelgrep tracks 62 OpenAI models with live benchmarks, speed, latency and per-provider pricing, led on intelligence by GPT-5.4. 25 of them qualify for this ranking.

More OpenAI rankings

All rankings