mistral
Mistral Medium 3.1
Mistral Medium 3.1 is Mistral's mid-tier multimodal model, accepting text and image inputs to produce text. At $0.40/M input and $2.00/M output, it is one of the more cost-efficient options in its performance bracket. In our 12-test benchmark suite, it ranks 15th out of 52 active models — a strong position for its price point. It achieves top-tier scores in constrained rewriting (5/5, tied for 1st with 4 other models), agentic planning (5/5, tied for 1st with 14 others), strategic analysis (5/5), and multilingual tasks (5/5). The 131,072 token context window is substantial, covering most real-world use cases comfortably. Among bracket peers, it sits at $2.00/M output alongside GPT-5 Mini and Devstral 2 2512 — both with different performance profiles.
Performance
Mistral Medium 3.1 ranks 15th out of 52 models overall in our benchmarks. Its top strengths: constrained rewriting (5/5, tied for 1st with 4 other models out of 53 tested), agentic planning (5/5, tied for 1st with 14 others out of 54), and strategic analysis (5/5, tied for 1st with 25 others out of 54). Multilingual (5/5), persona consistency (5/5), and long context (5/5) also scored at the top tier. These are particularly strong results for editorial, analytical, and multi-step planning workflows. Weaker areas include creative problem solving (3/5, rank 30 of 54) and safety calibration (2/5, rank 12 of 55 — at the field median). Faithfulness came in at 4/5 but ranked only 34th of 55, indicating mid-pack performance on adherence to source material.
Pricing
Mistral Medium 3.1 costs $0.40 per million input tokens and $2.00 per million output tokens. At 10 million output tokens per month, that is $20. At 100 million tokens, $200. Among bracket peers at similar performance levels, GPT-5 Mini and Devstral 2 2512 also cost $2.00/M output. Mistral Medium 3.1 offers image support on top of text at this price, which adds value for document processing workflows. Compared to higher-performing models like Claude Sonnet 4.6 at $15/M output or GPT-5.2 at $14/M, Mistral Medium 3.1 delivers competitive benchmark scores at a fraction of the cost. For teams running high-output workloads — 50 to 100 million tokens monthly — the $2.00/M rate keeps costs in the $100-200 range.
mistral
Mistral Medium 3.1
Benchmark Scores
External Benchmarks
Pricing
Input
$0.400/MTok
Output
$2.00/MTok
modelpicker.net
Real-World Costs
Pricing vs Performance
Output cost per million tokens (log scale) vs average score across our 12 internal benchmarks
Try It
from openai import OpenAI
client = OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key="YOUR_OPENROUTER_KEY",
)
response = client.chat.completions.create(
model="mistralai/mistral-medium-3.1",
messages=[
{"role": "user", "content": "Hello, Mistral Medium 3.1!"}
],
)
print(response.choices[0].message.content)Recommendation
Mistral Medium 3.1 is well-suited for constrained writing tasks (5/5, one of the top 5 models), strategic analysis, agentic planning pipelines, and multilingual content workflows. The image input support adds value for document processing and visual content tasks. At rank 15 of 52 and $2.00/M output, it offers a strong performance-to-cost ratio for teams that don't need frontier-tier pricing. Look elsewhere if creative problem solving is primary (3/5, rank 30/54) or if safety calibration in high-risk scenarios is critical (2/5, at the field median). For straightforward text-only workloads where creative output matters, models with higher creative problem solving scores may be a better fit.
How We Test
We test every model against our 12-benchmark suite covering tool calling, agentic planning, creative problem solving, safety calibration, and more. Each test is scored 1–5 by an LLM judge. Read our full methodology.