MODELS
o1
OpenAI's first reasoning model — the original 'thinks before answering' release.
Specs
- Context window
- 200,000
- Max output
- 100,000
- Modalities
- text, image
- Tool use
- —
- Vision
- ✓
- Streaming
- —
- License
- proprietary
- Released
- 2024-12-05
Pricing
- Input / 1M
- $15.00
- Output / 1M
- $60.00
- Cached input / 1M
- $7.50
Cost estimate
o1 was OpenAI's December 2024 reasoning model — the first widely-released LLM that visibly 'thinks' before answering, generating long internal chains of thought before the final response. Strong on math, code, and PhD-level science benchmarks. No tool use, no streaming, no system messages, $15/$60. Released alongside o1-mini and a rare o1-pro tier.
Editor's verdict
Historic 'first reasoning model' status — but already overtaken by o3 and o3-mini at every price point. Hidden chain-of-thought is opaque (you pay for tokens you can't see), and the no-tool-use limitation hurts agentic flows. Pick this only if you're benchmarking the original reasoning paradigm; for production reasoning workloads in 2026, o3 or DeepSeek R1 win.
Reviews
No reviews yet. Be the first.
Last updated: 2026-04-29