Best for high-volume agentic work on a tight budget.
- 1M context
- Text only
- Input
- $0.14 / 1M
- Cache read
- $0.028 / 1M
- Output
- $0.28 / 1M
- Intelligence
- 34
- DeepSWE
- 54.4%* reported by the publisher
- Terminal-Bench
- 78.7%
Models
Pick the right model.
Every model we serve, side by side: list prices, context windows, and scores on the benchmarks we trust for agentic coding.
Best for fast agentic coding at a low price.
- 1M context
- Vision
- Input
- $0.15 / 1M
- Cache read
- $0.028 / 1M
- Output
- $0.60 / 1M
- Intelligence
- 39
- DeepSWE
- 74.2%* reported by the publisher
- Terminal-Bench
- 90.6%* reported by the publisher
Best for complex, long-horizon coding.
- 1M context
- Text only
- Input
- $1.40 / 1M
- Cache read
- $0.26 / 1M
- Output
- $4.40 / 1M
- Intelligence
- 45
- DeepSWE
- 69%
- Terminal-Bench
- 83.9%
Best for the hardest, highest-stakes tasks.
- 1M context
- Vision
- Input
- $3.00 / 1M
- Cache read
- $0.30 / 1M
- Output
- $15.00 / 1M
- Intelligence
- 44
- DeepSWE
- 68.5%
- Terminal-Bench
- 85%
Best for everyday coding at a flash price.
- 1M context
- Vision
- Input
- $0.15 / 1M
- Cache read
- $0.03 / 1M
- Output
- $0.50 / 1M
- Intelligence
- 42
- DeepSWE
- 63.4%
- Terminal-Bench
- 84.3%
Best for fast side tasks: scouting, summaries.
- 256K context
- Vision
- Input
- $0.15 / 1M
- Cache read
- $0.05 / 1M
- Output
- $1.00 / 1M
- Intelligence
- 18
- DeepSWE
- n/a
- Terminal-Bench
- 44.9%
About the scores
Scores as of October 3, 2026. Each model's page links every score to its source.
Artificial Analysis Intelligence Index
By Artificial Analysis
An independent composite of reasoning, knowledge, maths, coding and agentic evaluations, run by Artificial Analysis on every model the same way.
MethodologyDeepSWE
By Datacurve
Long-horizon software engineering: real repository tasks an agent has to finish end to end.
MethodologyTerminal-Bench 2.1
By tbench.ai
Hard, realistic tasks an agent completes in a terminal sandbox: building, debugging, configuring.
Methodology* Reported by the model's publisher on its model card. Publishers run their own harnesses, so these compare less cleanly than the independent measurements, which carry no mark. No number here is estimated: a model without one has not been measured on that benchmark yet.
One key, every model.
Create a key once, then switch models by changing one string in your tool.
Looking for a model we no longer serve? See past models.