token.app › Compare › Llama 3.3 70B Instruct vs gpt-oss-20b

Llama 3.3 70B Instruct vs gpt-oss-20b

gpt-oss-20b leads on 2 of 2 shared benchmarks. gpt-oss-20b is 4.3× cheaper per token on a blended 3:1 basis.

Head to head
Blended price uses a 3:1 input:output mix
Llama 3.3 70B Instructgpt-oss-20b
ProviderMetaOpenAI
Input $/1M$0.100$0.018
Output $/1M$0.320$0.090
Blended $/1M$0.155$0.036
Context window131K131K
Max output16K33K
ReleasedDec 2024Aug 2025
LicenceOpen weightsOpen weights
GPQA Diamond 47.4% 60.8%
OTIS Mock AIME 5.1% 65.3%
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose Llama 3.3 70B Instruct if…
  • No clear advantage on the data we hold.
Choose gpt-oss-20b if…
  • Costs less per token — $0.036 vs $0.155 blended
  • Higher GPQA Diamond (60.8% vs 47.4%)
  • Higher OTIS Mock AIME (65.3% vs 5.1%)
  • Newer model (released Aug 2025)
FAQ
Which is better, Llama 3.3 70B Instruct or gpt-oss-20b?
Across the 2 benchmarks both models have independently-run scores for, Llama 3.3 70B Instruct leads on 0 and gpt-oss-20b leads on 2. Benchmarks are one input — price and context window are on this page too.
Is Llama 3.3 70B Instruct cheaper than gpt-oss-20b?
gpt-oss-20b is cheaper: $0.036 versus $0.155 per million tokens blended at 3:1 input:output. Input alone: $0.100 vs $0.018. Output alone: $0.320 vs $0.090.
What context windows do Llama 3.3 70B Instruct and gpt-oss-20b have?
Llama 3.3 70B Instruct: 131,072 tokens. gpt-oss-20b: 131,072 tokens.
Who makes Llama 3.3 70B Instruct and gpt-oss-20b?
Llama 3.3 70B Instruct is made by Meta. gpt-oss-20b is made by OpenAI.
Related comparisons