GPT OSS 20B
GPT OSS 20BAlternatives?
CheaperGPT-5.6 Lunanon-thinking+1-12%$0.01BetterGPT-6 Lunamedium+2.5+53%$0.017Function calling only with none · Chat Completions+3
OpenQwen3.8 27Bxhigh+5.6+8,698%$1.01Memory (4-bit): 13 GB → 17 GB+5
CheaperGranite 4.2 3BthinkingOpensame-48%$0.006Context: 131K → 128K
CheaperMinistral 3 3B 2512Opensame-31%$0.008GPT-5.6 Lunanon-thinking+1-12%$0.01GPT-6 Lunalowsame-61%$0.005Function calling only with none · Chat Completions+3Granite 4.2 3BthinkingOpensame-48%$0.006Context: 131K → 128KMinistral 3 3B 2512Opensame-31%$0.008
Gemma 4 E4BthinkingOpensame-29%$0.008Context: 131K → 128KGPT-6 Lunamedium+2.5+53%$0.017Function calling only with none · Chat Completions+3
Claude Haiku 5.5low+12.6+114%$0.024Remove sampling parameters+5GPT-5.6 Lunahigh+2.5+284%$0.044
MiMo v2.6 FlashthinkingOpen+22.7+445%$0.062Memory (4-bit): 13 GB → 186 GB+4GPT-6.1 Sollow+30.8+1,042%$0.13Remove unsupported parameters+5Qwen3.8 27BxhighOpen+5.6+8,698%$1.01Memory (4-bit): 13 GB → 17 GB+5IK2 Horizon 7BthinkingOpen+1+1,358%$0.30 / $1.20
North Mini CodeOpensame——Memory (4-bit): 13 GB → 18 GB
Muse Glimmer 30BhighOpensame+403%$0.058Memory (4-bit): 13 GB → 18 GB
Nemotron 3.5 LightningthinkingOpensame+594%$0.079Memory (4-bit): 13 GB → 20 GBGPT-6 Astrahigh+54+14,966%$1.73Monitoring can return HTTP 403+2GPT-6.1 Solhigh+51.5+2,687%$0.32Remove unsupported parameters+5GPT-6 Solhigh+26.3+3,156%$0.37GPT-5.6 Solhigh+20.7+6,956%$0.81GPT-5.5 Instant (June 2026)+12.6+5,938%$0.69GPT-5.5high+9.1+13,359%$1.54Prompts with >272K input tokensGPT-6 Lunahigh+4.5+154%$0.029Function calling only with none · Chat Completions+3GPT-5.6 Lunahigh+2.5+284%$0.044All OpenAI models →Run it yourself
estimateFAQ
How much does GPT OSS 20B cost?›
$0.018 per 1M input tokens and $0.09 per 1M output tokens (cheapest host: Darkbloom). A typical benchmark task costs about $0.011 at high effort.
Is GPT OSS 20B free to use?›
There is no free API; the cheapest paid price is $0.018 per 1M input tokens. The weights are open, so you can run it yourself.
How good is GPT OSS 20B, and what is it best at?›
#174 of 334 in Chat & Knowledge, #75 of 132 in Coding, #131 of 133 in Agents & Tool Use, #165 of 195 in Writing & Language, #106 of 133 in Vision & Documents.
Is GPT OSS 20B better than GPT OSS 120B?›
On Terminal-Bench 4.0, GPT OSS 20B scores 0 vs 0 — about the same. It is 88% cheaper per task.
How can I use GPT OSS 20B?›
Through the API of AWS Bedrock, AkashML, Amazon Bedrock, CoreWeave, Darkbloom and 7 more.
What is the context window of GPT OSS 20B?›
131K tokens.
When was GPT OSS 20B released? Is there a newer version?›
Released 2025-08-05. It is the current version in its line.
Is GPT OSS 20B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run GPT OSS 20B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can GPT OSS 20B read images?›
No, it accepts text only.
What is GPT OSS 20B?›
GPT OSS 20B is an open-weights language model from OpenAI, released Aug 2025.
What are the best alternatives to GPT OSS 20B?›
GPT-5.6 Luna (cheaper), GPT-6 Luna (better), Qwen3.8 27B (open weights).