OpenAI's most powerful reasoning model. Top scores on AIME, GPQA, and frontier coding benchmarks.
Loading...
Loading...
Extended chain-of-thought models that "think before they answer" — achieving frontier performance on math, coding, and science benchmarks. Compare API pricing, provider availability, and context windows across all major reasoning models.
Reasoning models spend extra compute thinking through a problem before producing a final answer. They generate internal chain-of-thought tokens (sometimes visible, sometimes hidden) which dramatically improves accuracy on problems requiring multi-step logic — STEM, proofs, complex code. The trade-off: they're typically 3–10× slower and more expensive than standard models. Use them when correctness matters more than speed.
OpenAI's most powerful reasoning model. Top scores on AIME, GPQA, and frontier coding benchmarks.
Compact reasoning model with tool use and vision at a fraction of o3 cost.
Efficient STEM reasoning at significantly lower cost than full o3.
OpenAI's original reasoning model with internal chain-of-thought for complex STEM tasks.
Anthropic's most capable model with hybrid extended thinking and 1M context.
Production-grade thinking model balancing cost and extended reasoning capability.
Frontier agentic coding and reasoning with extended thinking mode.
Google flagship with 2M context and deep-think mode for multi-step scientific reasoning.
Google's best price-to-performance thinking model with 1M token context.
Open-source reasoning model matching frontier closed-source models at 90% lower cost.
Open reasoning pioneer — first open model to match o1 performance. Available on 15+ providers.
xAI's frontier reasoning model with real-time data access and strong STEM benchmarks.
Grok 3 with Think mode — top-ranked on AIME 2025 and GPQA Diamond at launch.
Open reasoning model matching o1/R1 at compact 32B size. Available on 10+ providers.
List order reflects available data only. Coverage is incomplete — not all models are indexed and benchmark scores may be missing, outdated, or sourced from different test conditions. This is not an endorsement or a definitive ranking. How we collect data →
Not sure which reasoning model fits your budget?
Use the comparison tool to see cost vs. benchmark trade-offs side-by-side.
Pricing updated daily via automated pipelines. How we collect data →