gpt-oss-20b is 3.7× cheaper per million tokens (blended). Llama 4 Scout has the larger context window (327,680 vs 131,072 tokens, 2.5×).
Llama 4 Scout vs gpt-oss-20b
Live catalog fields. Best-in-row is highlighted.
Llama 4 Scoutopen
Meta · meta-llama__llama-4-scout
gpt-oss-20bopen
OpenAI · openai__gpt-oss-20b
Identity
| Field | meta-llama__llama-4-scout | openai__gpt-oss-20b |
|---|---|---|
| Provider | Meta | OpenAI |
| Slug | meta-llama__llama-4-scout | openai__gpt-oss-20b |
| Status | live | live |
| Open weights | Yes | Yes |
| License | — | — |
| HuggingFace | meta-llama/Llama-4-Scout-17B-16E-Instruct | openai/gpt-oss-20b |
Quality
| Field | meta-llama__llama-4-scout | openai__gpt-oss-20b |
|---|---|---|
| Quality index (0–100) | — | — |
Cost
| Field | meta-llama__llama-4-scout | openai__gpt-oss-20b |
|---|---|---|
| Input $ / 1M tokens | $0.10 | $0.02 |
| Cached input $ / 1M | — | $0.01 |
| Output $ / 1M tokens | $0.30 | $0.09 |
Context
| Field | meta-llama__llama-4-scout | openai__gpt-oss-20b |
|---|---|---|
| Max context tokens | 327,680 | 131,072 |
| Max output tokens | 16,384 | 32,768 |
Modalities
| Field | meta-llama__llama-4-scout | openai__gpt-oss-20b |
|---|---|---|
| Input | text, image | text |
| Output | text | text |