gpt-oss-20b
by OpenAItext->text14 endpoints
Neura Intelligence Index
30.9/ 100
Rank
#261—
Confidence
lowBenchmarks
3(8% coverage)
Overview
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...
Capabilities
Text generation
Tool use / Function calling
Structured output
Extended reasoning
Prompt caching
Long context
Top-P sampling
Stop sequences
Deterministic seed
Log probabilities
Modalities
Input
Text
Output
Text
Technical Specifications
Context Window
131.1K tokens
Max Output
118.0K tokens
Knowledge Cutoff
2024-06-30
Tokenizer
GPT
Uptime (24h)
10000.0%
Parameters
20900000000B
License
apache_2_0Open Weights
Country
🇺🇸US
Release Date
August 5, 20251y
Supported Parameters
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Benchmark Performance
reasoning
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 71.5 | verified |
| Humanity's Last Exam | 10.9 | verified |
agent
| Benchmark | Score | Source |
|---|---|---|
| TAU-Bench Retail | 54.8 | verified |
Performance Over Time
Pricing
Per 1M tokens
Input$0.03
Output$0.13
Cache read$0.03
Added
August 5, 2025
Last synced 9/16/2026