Codestral 25.08 alternatives
Same input types and capabilities, at least half the context window · USD per 1M tokens
We found 10 models that can replace Codestral 25.08 ($0.30 input / $0.90 output per million tokens). 10 of them are cheaper. The cheapest, Ministral 3 3B, costs 77.8% less on a blended 3:1 basis: $1.20 instead of $4.80 per month for a chatbot workload (10M input / 2M output tokens).
// 10 alternatives
| model | lab | in $/1M | out $/1M | ctx | vs Codestral 25.08 | chatbot / mo |
|---|---|---|---|---|---|---|
| Ministral 3 3B | Mistral | $0.10 | $0.10 | 131.1K | −77.8% | $1.20 |
| Muse Spark 1.2 Contributor | Meta | $0.10 | $0.20 | 1.05M | −72.2% | $1.40 |
| Muse Spark 1.3 Contributor | Meta | $0.10 | $0.20 | 1.05M | −72.2% | $1.40 |
| Ministral 3 8B | Mistral | $0.15 | $0.15 | 262.1K | −66.7% | $1.80 |
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | 1.05M | −61.1% | $1.80 | |
| Claude Haiku 5.5 · compare | Anthropic | $0.10 | $0.50 | 1M | −55.6% | $2.00 |
| Ministral 3 14B | Mistral | $0.20 | $0.20 | 262.1K | −55.6% | $2.40 |
| GPT-6 Luna · compare | OpenAI | $0.10 | $0.50 | 1.05M | −55.6% | $2.00 |
| Mistral Small 4 | Mistral | $0.15 | $0.60 | 262.1K | −41.7% | $2.70 |
| GPT-4o mini | OpenAI | $0.15 | $0.60 | 128K | −41.7% | $2.70 |
// how we pick alternatives
An alternative must accept every input type Codestral 25.08 accepts (text, pdf), offer the same tool-calling and reasoning support, and have at least half of its 256K-token context window. Models are sorted by blended price (3 × input + output) / 4, using base-tier list prices refreshed daily.