Skip to main content

GPT OSS 20B

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.

Model card on Hugging Face
Input / output price
$0.018 / $0.09 per 1M
Context
131K tokens
Released
Aug 5, 2025
Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "gpt-oss-20b".

Read docs

Providers

9 providers, cheapest first, OpenRouter last, priced per 1M tokens after gateway discounts.

ProviderInputOutput
Concentrateconcentrate/gpt-oss-20b
$0.03/M$0.14/M
DeepInfradeepinfra/gpt-oss-20b
$0.03/M$0.14/M
Parasailparasail/gpt-oss-20b
$0.03/M$0.15/M
Featherlessfeatherless/gpt-oss-20b
$0.04/M$0.15/M
IO.netionet/gpt-oss-20b
$0.055/M$0.176/M

Is a provider down? AWS Bedrock status, Groq status, OpenRouter status

Pricing

Averaged across 9 providers, in USD per 1M tokens after gateway discounts.

Average input
$0.06 per 1M
Average output
$0.19 per 1M
Average cache read
$0.11 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$0.20/MPrime IntellectLowest$0.018/MOpenRouter

Throughput and latency

30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.

Output speed

Higher is better

60tokens/sacross providers

Faster than 69% of the other 596 models on the gateway

Time to first token

Lower is better

0.9sacross providers

Faster than 76% of the other 153 models on the gateway

Response latency

Lower is better

24sacross providers

Faster than 8% of the other 768 models on the gateway

Benchmarks

The Respan Index plus 8 published scores from 5 sources, each linked to where it was published.

13.6/ 100

Respan Index#166 of 176

176 ranked models. The shaded band is this model's 95% interval, 10.2 to 21.5. Qwen3 30B-A3B Instruct 2507 and DeepSeek V3 fall inside it, so the gap is within the margin of error.
GPT OSS 20B by category: Coding 4.7, Agents 7.8 (estimated), Reasoning 18.5, Knowledge & language 7.1 (estimated), Human preference 34.5
  1. 166gpt-oss-20b13.6
  2. 165Qwen3 30B-A3B Instruct 250713.8
  3. 167DeepSeek V313.2

Published scores

BenchmarkScore
SWE-bench VerifiedOpenAI60.7%
WeirdMLHåvard Tveit Ihle40.9%

Compare GPT OSS 20B: gpt-oss-20b vs gpt-oss-120b, gpt-oss-20b vs GPT-6.2 Astra, gpt-oss-20b vs GPT-6 Astra, gpt-oss-20b vs GPT-6.1 Sol, gpt-oss-20b vs GPT-6.1 Terra, gpt-oss-20b vs GPT-6.2 Sol

Apps

Use GPT OSS 20B in the tools you already run. Each guide points the app at the Respan gateway.

Setup guides for more apps are in the integration docs.

GPT OSS 20B FAQ

Short answers, built from the data on this page.

GPT OSS 20B is a model from OpenAI, released on Aug 5, 2025. 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.

On the Respan gateway, GPT OSS 20B costs from $0.018 per 1M input tokens and $0.09 per 1M output tokens on OpenRouter, after gateway discounts. Averaged across its 9 providers, it costs $0.06 per 1M input tokens and $0.19 per 1M output tokens.

GPT OSS 20B is available through 9 providers: Concentrate, DeepInfra, Parasail, Featherless, IO.net, Bedrock, and 3 more.

GPT OSS 20B accepts up to 131K tokens of context on its largest provider route.

Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "gpt-oss-20b". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

GPT OSS 20B scores 13.6 out of 100 on the Respan Index v1.1, ranking #166 of 176 models. By category: Coding 4.7, Agents 7.8 (estimated), Reasoning 18.5, Knowledge & language 7.1 (estimated), and Human preference 34.5. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.

GPT OSS 20B has 8 published benchmark scores, including SWE-bench Verified 60.7%, GPQA Diamond 71.5%, and AIME 2025 89.2%. Each score links to where it was published; the data is collected by Respan.

Certified where it counts.

Built to meet the security and privacy standards that enterprise and healthcare teams require.

ISO 27001

The internationally recognized standard for information security management.

SOC 2

Secure, compliant management of your data across all of our systems.

GDPR

Operated under GDPR, the world's strictest standard for data privacy.

HIPAA

HIPAA compliant, with a BAA available for healthcare teams.

Works with your entire stack

Use Respan with your favorite frameworks and tools.

AnthropicAI21 LabsxAIVercel AI SDK
LangChainOpenRouterAssemblyAICursor
LiteLLMGroqBasetenMem0
CohereCogneeLlamaIndexFireworks
DeepSeekLinkupHaystackTogether AI
PerplexityInference.netPostHogRubyLLM
SlackMastraAzure OpenAIMistral
AWS BedrockMoonshotHyperspellBAML
ModaInstructorGoogle GeminiParasail
ReductoZapierAgnoNebius
OpenAINovita AIReplicateSuperagent

Built for AI agents.
Break less.
Ship more.