GPT-5.4 vs Mistral Large 3: API Pricing
Input and output token rates, context windows, and real monthly cost for GPT-5.4 (OpenAI) and Mistral Large 3 (Mistral), side by side. Prices are standard on-demand rates as of June 2026.
The short answer
For a typical coding-agent workload (60M in / 12M out per month), Mistral Large 3 is the cheaper option at $48.00/mo versus $330/mo for GPT-5.4 - about 85% less (6.9× cheaper). On the headline sticker of 1M input + 1M output, GPT-5.4 is $17.50 and Mistral Large 3 is $2.00.
Rates at a glance
| GPT-5.4 | Mistral Large 3 | |
|---|---|---|
| Input ($/1M tokens) | $2.50 | $0.50 |
| Output ($/1M tokens) | $15.00 | $1.50 |
| Blended (1M in + 1M out) | $17.50 | $2.00 |
| Context window | 1,050K | 128K |
| Type | Proprietary | Open-weight |
| Provider | OpenAI | Mistral |
Monthly cost by workload
Estimated monthly API spend at each workload's token volume. Output usually costs several times input, so the winner can flip with your mix.
| Workload | GPT-5.4 | Mistral Large 3 | Cheaper |
|---|---|---|---|
| Chatbot / assistant10M in / 3M out | $70.00/mo | $9.50/mo | Mistral Large 3 |
| Coding agent60M in / 12M out | $330/mo | $48.00/mo | Mistral Large 3 |
| RAG / summarization40M in / 4M out | $160/mo | $26.00/mo | Mistral Large 3 |
| Batch / classification20M in / 2M out | $80.00/mo | $13.00/mo | Mistral Large 3 |
Want your own in/out split? Use the full interactive comparator to rank every model and provider for your exact workload.
Frequently asked questions
Is GPT-5.4 or Mistral Large 3 cheaper?
It depends on your input/output mix, but for a typical coding-agent workload (60M in / 12M out per month) Mistral Large 3 costs $48.00/mo versus $330/mo for GPT-5.4 - about 85% less (6.9x). On the headline sticker (1M input + 1M output), GPT-5.4 is $17.50 and Mistral Large 3 is $2.00.
What are the token rates for GPT-5.4 and Mistral Large 3?
GPT-5.4 (OpenAI) is $2.50 per 1M input and $15.00 per 1M output. Mistral Large 3 (Mistral) is $0.50 per 1M input and $1.50 per 1M output. These are standard on-demand rates, not cached or batch.
Is GPT-5.4 or Mistral Large 3 open-weight?
GPT-5.4 is proprietary and Mistral Large 3 is open-weight. Open-weight models can be self-hosted or run on third-party hosts at different rates, so the first-party price shown here is a starting point, not the only option.
What context window do GPT-5.4 and Mistral Large 3 support?
GPT-5.4 supports 1,050K tokens and Mistral Large 3 supports 128K tokens. Some models also step up pricing past a size threshold - check the source pricing pages for long-context tiers.
More pricing comparisons
Stay ahead of the AI tools curve
Picks, reviews, and automation tips every weekday. Free, no spam.