OpenAI o1
Inference-Time Reasoning
OpenAI's first model trained explicitly to "think before it answers" — generating a long chain of internal reasoning before producing its final response. On math olympiad and competitive-programming problems, it crossed thresholds (top human percentiles) that prior models couldn't approach. Trade-off: each query takes 10-60 seconds and costs 5-50× more than GPT-4o.
Intelligence
Medium
Speed
Medium
89 tok/s output
Cost
High
$16.50 in / $66.00 out
Context
128K
Up to 128,000 tokens
How are Intelligence, Speed & Cost bucketed?
Intelligence and Speed buckets are percentile ranks on
Artificial Analysis. Cost buckets are fixed dollar
thresholds keyed off output-token price ($/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
o1 ended the "scaling laws determine everything" era. After o1, capability could be bought at inference time, not just training time. This is the most important technical pivot in the LLM landscape since GPT-3.
Core Capabilities
Long Documents
Handles entire codebases, books, and multi-doc RAG.
Generative
Produces images, video, audio, or other media.
Research
Foundational paper or scientific contribution.
Context Window
128k tokens
≈ 98 pages
4k Chat 聊天
32k Long docs 长文档
128k This model 本模型
400k Multi-doc 多文档
1M Codebase 整个代码库
10M
Availability
API
Available
Product / App
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
Context / memory
Context window size · log-scaled
6.0
9.0
6.0
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
- Preferred to GPT-4o by a large margin in data analysis, coding, and math — but not preferred on natural-language tasks
- Genuinely novel: chain-of-thought is internalised, not prompted — model recognises and corrects its own mistakes
- Slow — minutes for hard problems vs seconds for GPT-4o; not for chat
- Hidden chain of thought is policy-restricted; users can lose access for trying to extract it
- Apple's GSM-Symbolic study showed 17.5% perf drop with renamed variables — fragility lurks under the high benchmark scores
- Token costs are heavy because of the reasoning trace, even when not shown to the user
Best use cases
- Olympiad-level math, physics, and chemistry problems
- Code that needs deep reasoning about correctness rather than fluency
- Strategic / multi-step planning where you can wait minutes for an answer
- Research workflows that benefit from explicit step-by-step decomposition
Tools to try
Not ideal for
- Real-time chat or low-latency tools — GPT-4o is faster
- Open-ended writing, creative copy, or natural-language tasks
- Cost-sensitive bulk inference — reasoning tokens are expensive
- Use cases where you need to inspect or steer the chain of thought
Model Evolution
o is OpenAI's language model family.