GPT-4o vs Mistral Large 3
Pricing, benchmarks, and use case comparison
Newer versions are out: GPT-4o has been replaced by GPT-6 Luna; Mistral Large 3 has been replaced by Mistral Large 4.
Verdict
Our pick: Mistral Large 3Both are deprecated: OpenAI removes the remaining GPT-4o API snapshots on January 14, 2027, and Mistral retires Mistral Large 3 on April 30, 2027. Between them, Mistral Large is the better pick for reasoning, multilingual work, and open weights you can keep self-hosting; for new projects, use GPT-6 Luna or Mistral Large 4.
Specs comparison
| GPT-4o | Mistral Large 3 | |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Type | Closed source | Open source |
| Context window | 128,000 tokens (16,384 max output) | ✓256K |
| Input / 1M tokens | $2.50 | ✓$0.50 |
| Output / 1M tokens | $10.00 | $1.50 |
| Release date | 2024-05 | 2025-12 |
Benchmarks
| Benchmark | GPT-4o | Mistral Large 3 |
|---|---|---|
| MMLU | 88.7% | - |
| HumanEval | 90.2% | - |
| MATH | 76.6% | - |
Scores sourced from official provider release posts and independent benchmark aggregators.
Both have named replacements: OpenAI recommends GPT-6 Luna in place of GPT-4o, and Mistral recommends Mistral Large 4 in place of Mistral Large 3.
Capability and benchmarks
Mistral Large is the stronger reasoner and coder on our capability index (reasoning 85 and coding 82, vs GPT-4o's 68 and 72). The published Mistral scores we hold are from Mistral Large 2, which reported 92.0% HumanEval and 84.0% MMLU. GPT-4o reports 90.2% HumanEval, 88.7% MMLU, and 76.6% MATH, and is tuned for fast, direct instruction-following. For nuanced reasoning, multilingual generation, and structured output, Mistral Large leads.
Status, price, and deployment
OpenAI deprecated its GPT-4o API snapshots on October 14, 2026: gpt-4o-2024-08-06 and gpt-4o-2024-11-20 are removed from the API on January 14, 2027 (the older gpt-4o-2024-05-13 snapshot shut down on October 23, 2026), and GPT-4o left ChatGPT on February 13, 2026. Until the removal it costs $2.50 input / $10 output per 1M (cached input $1.25) with a 128K context and text and image input. Mistral Large 3 (mistral-large-2512) was deprecated on October 31, 2026 and retires from Mistral's API on April 30, 2027. It is open-weight under Apache 2.0 with a 256K context, and Mistral's pricing page lists Mistral Large at $0.50/$1.50 per 1M, below GPT-4o on both input and output.
Which to pick
- Pick Mistral Large for stronger reasoning and coding, multilingual and structured-output work, lower cost, and Apache 2.0 weights you can keep running after the API retirement.
- Pick GPT-4o only to keep an existing integration running until January 14, 2027; OpenAI's recommended replacement, GPT-6 Luna, costs $0.10/$0.50 per 1M with a 1.05M context.
- For new Mistral projects, Mistral Large 4 costs $0.70/$2.10 per 1M with a 1M context and open weights.
Which should you choose?
Choose GPT-4o if...
- →Everyday assistant, drafting, summarization, and classification tasks
- →Latency- and cost-sensitive applications at scale
- →Multimodal tasks needing image understanding with fast responses
Choose Mistral Large 3 if...
- →You already run Mistral Large 3 and need time to migrate before April 30, 2027
- →You want to self-host an Apache 2.0 model that is smaller than Mistral Large 4
- →You want the slightly lower API price ($0.50/$1.50 vs $0.70/$2.10 per 1M tokens)
- →You need the option to self-host for data sovereignty