Current as of March 2026. O1 is OpenAI’s reasoning model: before it writes a word of output, it works through the problem internally. That hidden reasoning phase is why it’s slower and more expensive than chat models — and also why it handles problems that make GPT-4o stumble. At $15/$60 per million tokens, it’s not for casual use.

Specs

ProviderOpenAI
Input cost$15 / M tokens
Output cost$60 / M tokens
Context window200K tokens
Max output100K tokens
ParametersN/A
Featuresfunction_calling, vision, reasoning

What it’s good at

Hard Reasoning Problems

Multi-step proofs, symbolic math, complex logic chains — this is what it’s built for. If GPT-4o keeps getting the answer wrong, O1 often gets it right.

Long Output

100K max output tokens. You can generate substantial code or documentation in a single pass.

Where it falls short

Latency

The internal reasoning phase can take seconds to minutes before the first output token. Users watching a blank screen will assume something broke.

Cost

$60/M output. High-volume agent loops will get expensive fast. Be selective about what you send here.

Best use cases with OpenClaw

  • Architectural Refactoring — Feed a 200K context of codebase and let it reason through dependency changes. The reasoning quality justifies the cost on hard problems.
  • Scientific and Mathematical Analysis — Dense formulas, logical inconsistencies in research papers, proofs. This is where O1 earns its price.

Not ideal for

  • Basic Summarization — Wasteful. GPT-4o-mini handles this for pennies.
  • User-facing Chat — The latency alone will kill the experience.

Run it through Haimaker

Skip juggling API keys. One Haimaker key gives you access to every model on the platform. Tell OpenClaw:

Add Haimaker as a custom provider to my OpenClaw config. Use these details:

- Provider name: haimaker
- Base URL: https://api.haimaker.ai/v1
- API key: [PASTE YOUR HAIMAKER API KEY HERE]
- API type: openai-completions

Add the auto-router model:
- haimaker/auto (reasoning: false, context: 128000, max tokens: 32000)

Create an alias "auto" for easy switching. Apply the config when done.

Or skip model selection entirely — Haimaker’s auto-router picks the best model for each task so you don’t have to.

OpenClaw setup

OpenClaw handles O1 natively through the OpenAI provider. Simply export your OPENAI_API_KEY and set the model ID to openai/o1 in your agent settings.

export OPENAI_API_KEY="your-key-here"

That’s it. OpenClaw picks up OpenAI models automatically.

How it compares

  • vs Claude 3.5 Sonnet — Sonnet is much faster and better for everyday coding. O1 wins when the problem requires genuine reasoning depth, not just competent code generation.
  • vs DeepSeek-R1 — R1 gets close on reasoning benchmarks at a fraction of the price. If API reliability matters less to you, R1 is worth testing first.

Bottom line

O1 is a specialized reasoning engine. Use it for the hard problems — debugging subtle logic, architectural decisions, math. Route everything else to cheaper models.

TRY O1 ON HAIMAKER


For setup instructions, see our API key guide. For all available models, see the complete models guide.