LANGUAGE MODEL OpenAI Last updated:

GPT-5.5 NEW

Agentic “Do the Work” Model (Thinking / Pro)

GPT-5.5 is OpenAI’s “agent” model: it’s built to take a messy goal, turn it into steps, use tools, and keep iterating until the work is done. It’s most valuable when the task isn’t one prompt → one answer (coding changes, multi-source research, long documents, or workflows that need checking and follow‑ups), and less about casual chat. Expect higher cost and latency than lighter models in exchange for better persistence and tool use. GPT-5.5 是 OpenAI 主打“能把事做完”的 Agent 模型:它更擅长把一个复杂目标拆成步骤, 调用工具/接口,反复检查与迭代,直到交付结果。它在需要多轮推进的工作里更有价值(改代码、 多来源调研、长文档处理、需要自检与回访的流程),而不是单纯闲聊。代价是更高的成本与更慢的 响应,但换来更强的持续执行与工具使用能力。

Intelligence
Top 1%
Speed
Slow
81 tok/s output
Cost
High
$5.00 in / $30.00 out
Context
922K
Up to 922,000 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on Artificial Analysis. Cost buckets are fixed dollar thresholds keyed off output-token price ($/M out).
Intelligence
  • Top 1%≤ 1%
  • Top 5%≤ 5%
  • Top 10%≤ 10%
  • Good≤ 25%
  • Medium≤ 50%
  • Below avg> 50%
Speed
  • Top 1%≥ 345 tok/s
  • Top 5%≥ 237 tok/s
  • Top 10%≥ 196 tok/s
  • Good≥ 146 tok/s
  • Medium≥ 90 tok/s
  • Slow< 90 tok/s
Cost
  • Freeopen weights · self-host
  • Low< $1 / M out
  • Moderate$1–5 / M out
  • High≥ $5 / M out

Why it matters

It’s a step toward “LLM as operator”: planning + tool use + persistence, not just generation quality.

Core Capabilities

Long Documents
Handles entire codebases, books, and multi-doc RAG.
Multimodal
Combines text, vision, and audio in one model.
Generative
Produces images, video, audio, or other media.
Reasoning
Solves complex math, logic, and planning tasks.

Context Window

922k tokens
≈ 709+ pages
4k Chat 聊天
32k Long docs 长文档
128k Books 整本书
400k Multi-doc 多文档
1M This model 本模型
10M

Availability

API
Available
Product / App
Not available
Open Source
Not released
Enterprise
Contact sales

Pricing Model

Pay per token
Input and output billed separately.
Pay-per-token

Capability / Performance

Where this model sits relative to the middle 60% of models in the tree. All scores are 0–10 (higher is better).

Lower 20% Upper 80% This model
Reasoning
AA Intelligence Index · scaled to 10
1.7
5.6
8.6
Coding
SciCode · scaled to 10
1.8
4.3
5.6
Agentic tasks
Terminal-Bench Hard · scaled to 10
0.2
3.6
6.1
Context / memory
Context window size · log-scaled
6.0
9.0
8.9
Cost efficiency
Input price ($/M tokens) · cheaper scores higher
6.2
10.0
3.8
Lower 20% 20th percentile — 20% of models score below this This model Where the current model lands Upper 80% 80th percentile — only 20% of models score above this Percentile boundaries are computed across every model in the tree that reports the underlying benchmark for each capability.

What it feels like

  • Later 2026 GPT-5 refresh — incremental quality bump over 5.2.
  • Pro variant hits 93.2% on GPQA Diamond and 40.3% on FrontierMath Tier 1-3 — new state-of-the-art on hard math/science
  • 400K-token context window — biggest OpenAI window yet, designed for long-running agents
  • GDPval state-of-the-art — outperforms industry professionals on well-specified tasks across 44 occupations

Best use cases

  • Hardest math/science research where Pro/Thinking modes pull ahead of every competitor
  • Long-running professional agent workflows (400K context + GDPval-leading task quality)
  • Knowledge work spanning 44 occupations — spreadsheets, presentations, multi-step projects

Tools to try

Not ideal for

  • Casual chat or creative writing — Instant tier feels muted compared to GPT-5.1
  • Cost-sensitive bulk inference — 40% price hike vs 5.1 with no automatic upgrade win

Model Evolution

GPT is OpenAI's language model family.

View full evolution tree →