OpenAI: gpt-oss-20b
OpenAI's gpt-oss-20b is an exceptionally cost-effective model at $0.03/1M input tokens, with strong math and livecodebench results and solid agentic scores — a compelling choice for high-volume, cost-sensitive workflows despite modest overall reasoning capability.
Assessment date: March 12, 2026
Our methodology takes into account a range of factors including pricing, functionality, capabilities, benchmark performance, and real-world applicability. Rankings are reviewed and updated regularly as new models are released. Issues with our rankings? Contact us
gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for lower-latency inference and deployability on consumer or single-GPU hardware. The model is trained in OpenAI’s Harmony response format and supports reasoning level configuration, fine-tuning, and agentic capabilities including function calling, tool use, and structured outputs.
Capabilities
Architecture
| Modality | Text → Text |
| Tokenizer | GPT |
| Parameters | 20B |
Performance Indices
Source: Artificial Analysis
Benchmark Scores
Evaluations
Benchmark data from Artificial Analysis and Hugging Face
Model Information
Pricing
| Token Type | Cost per 1M tokens | Cost per 1K tokens |
|---|---|---|
| Input | $0.03 | $0.000030 |
| Output | $0.14 | $0.000140 |
Live Performance
Live endpoint metrics — refreshed every 30 minutes.
