GPT-3.5 Turbo 16k vs gpt-oss-20b: pricing comparison
gpt-oss-20b is the cheaper option at $0.02/$0.09 per 1M input/output tokens — GPT-3.5 Turbo 16k ($3/$4) costs about 167x more per input token. Full spec-by-spec breakdown below.
| GPT-3.5 Turbo 16k | gpt-oss-20b | |
|---|---|---|
| Input /1M tokens | $3 | $0.02 |
| Output /1M tokens | $4 | $0.09 |
| Cache read /1M | — | $0.01 |
| Context window | 16K | 131K |
| Provider | OpenAI | OpenAI |
| Vision input | No | No |
| Released | Aug 2023 | Aug 2025 |
Real workload costs: GPT-3.5 Turbo 16k vs gpt-oss-20b
Cost per single request at common token profiles — the cheapest model for each workload is highlighted.
| Workload | Tokens (in / out) | GPT-3.5 Turbo 16k | gpt-oss-20b |
|---|---|---|---|
| Chatbot message | 500 / 300 | $0.0027 | <$0.0001 |
| RAG query with context | 4,000 / 500 | $0.01 | $0.0001 |
| Document summarization | 20,000 / 1,000 | $0.06 | $0.0004 |
| Agent coding session | 100,000 / 5,000 | $0.32 | $0.0022 |
Frequently asked questions
Which is cheaper: GPT-3.5 Turbo 16k vs gpt-oss-20b?
gpt-oss-20b is the cheapest on input tokens at $0.02 per 1M, and gpt-oss-20b is cheapest on output at $0.09 per 1M. GPT-3.5 Turbo 16k costs about 167x more per input token than gpt-oss-20b.
How much does GPT-3.5 Turbo 16k cost per 1M tokens?
GPT-3.5 Turbo 16k costs $3 per 1M input tokens and $4 per 1M output tokens on the OpenAI API, with a 16K-token context window.
How much does gpt-oss-20b cost per 1M tokens?
gpt-oss-20b costs $0.02 per 1M input tokens and $0.09 per 1M output tokens on the OpenAI API, with a 131K-token context window.
What does a typical request cost on GPT-3.5 Turbo 16k vs gpt-oss-20b?
For a typical request with 2,000 input and 500 output tokens: GPT-3.5 Turbo 16k costs $0.0080, gpt-oss-20b costs <$0.0001. At 1,000 requests/day for a month that is $240 for GPT-3.5 Turbo 16k vs $2.43 for gpt-oss-20b.
Estimate your own workload with the per-model calculators (GPT-3.5 Turbo 16k, gpt-oss-20b), browse all current prices on the live model pricing dashboard, or run these models head-to-head on a real prompt (free account required). Prices may differ from provider list prices for batch or tiered usage.
Pricing data via the OpenRouter models API.