gpt-oss-20b is 7.5× cheaper per million tokens (blended). Codestral 2508 has the larger context window (256,000 vs 131,072 tokens, 2.0×). gpt-oss-20b is open-weights; the other is not.
Codestral 2508 vs gpt-oss-20b
Live catalog fields. Best-in-row is highlighted.
Codestral 2508
Mistral · mistralai__codestral-2508
gpt-oss-20bopen
OpenAI · openai__gpt-oss-20b
Identity
| Field | mistralai__codestral-2508 | openai__gpt-oss-20b |
|---|---|---|
| Provider | Mistral | OpenAI |
| Slug | mistralai__codestral-2508 | openai__gpt-oss-20b |
| Status | live | live |
| Open weights | No | Yes |
| License | — | — |
| HuggingFace | — | openai/gpt-oss-20b |
Quality
| Field | mistralai__codestral-2508 | openai__gpt-oss-20b |
|---|---|---|
| Quality index (0–100) | — | — |
Cost
| Field | mistralai__codestral-2508 | openai__gpt-oss-20b |
|---|---|---|
| Input $ / 1M tokens | $0.30 | $0.03 |
| Cached input $ / 1M | $0.03 | $0.03 |
| Output $ / 1M tokens | $0.90 | $0.13 |
Context
| Field | mistralai__codestral-2508 | openai__gpt-oss-20b |
|---|---|---|
| Max context tokens | 256,000 | 131,072 |
| Max output tokens | 204,800 | 117,964 |
Modalities
| Field | mistralai__codestral-2508 | openai__gpt-oss-20b |
|---|---|---|
| Input | text, file | text |
| Output | text | text |