Compare / Mistral Medium 3.1 vs Mistral Medium 3.5

Mistral Medium 3.1 vs Mistral Medium 3.5

Which is cheaper, which is more capable, and which does better at each kind of work, from 22 results on 3 sources that measured both models.

Mistral AIMistral Medium 3.1 Released 12 Aug 2025 · $0.40 in · $2.00 out
Mistral AIMistral Medium 3.5 Released 28 Apr 2026 · $1.50 in · $7.50 out
Mistral Medium 3.1 wins 13 too close to callMistral Medium 3.5 wins 18

Of the 22 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 16 November 2026 (catalogue release 2026-11-16-49bdfa045dee).

Mistral Medium 3.1 or Mistral Medium 3.5? The short answer

Mistral Medium 3.5 is the newer version and scores higher (Intelligence Index 14.2 against 9.2), but Mistral Medium 3.1 is cheaper (3.8 times less per token): upgrade if the gain is worth the price.

  • Mistral Medium 3.1 is cheaper: $0.80 against $3.00 per million tokens, 3.8 times less.
  • Mistral Medium 3.5 scores higher on the Artificial Analysis Intelligence Index: 14.2 against 9.2.
  • Mistral Medium 3.5 is newer: released 28 Apr 2026, against 12 Aug 2025 for Mistral Medium 3.1.
  • On third-party benchmarks, on the results both models have: Mistral Medium 3.5 does better for agents and tool use, reasoning and knowledge and writing.

Head to head

9.2
Intelligence Index
14.2
$0.80
Price per million tokens, blendedlower is better
$3.00
131,072
Context window, tokens
262,144

Every fact side by side

Mistral Medium 3.1Mistral Medium 3.5
Price per million tokens$0.40 in · $2.00 out$1.50 in · $7.50 out
Blended (3 in : 1 out)$0.80$3.00
Cheapest host, blended$0.80 (Mistral)$3.00 (Mistral)
Cached input, per million–$0.15 cached input
Intelligence Index9.214.2
Spring Prompt overall–15
Output speed, tokens a second–157
Context window131,072 tokens262,144 tokens
Released12 Aug 202528 Apr 2026 (newer)
WeightsClosedClosed
Developer based inFranceFrance

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Product listingsTurning a sparse product feed and photos into listings that can go live not measured 23 of 32
Decks from an analysisTurning a finished analysis into a deck you could present as it is not measured 31 of 32
User surveysPlanning a user survey and reading its results without being misled not measured 21 of 23
Marketing planningPlanning a year of ad spend without overspending not measured 32 of 32
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 128 of 159 100 of 159
Reasoning and knowledgeHard questions across science, maths and general knowledge 151 of 187 123 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 110 of 169 85 of 169
Sticking to the factsSummarising without inventing things, and factual answers 122 of 139 98 of 139
Speed under pressureGood decisions against a real clock (fast chess) not measured 10 of 35

From our research

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureMistral Medium 3.1Mistral Medium 3.5
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 130 · 99 of 387 measured 1,408 1,427
Overall · Text factualityArena rating, higher is better 99 · 97 of 163 measured 1,429 1,431
Business, management and finance · TextArena rating, higher is better 129 · 98 of 380 measured 1,407 1,430
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 187 · 139 of 249 measured 9.2 14.2
GPQA Diamond% of questions, higher is better 185 · 140 of 232 measured 58.8% 74.8%
Humanity's Last Exam% of questions, higher is better 204 · 139 of 250 measured 4.7% 13.8%
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 114 · 86 of 230 measured 39.5 45.5no reasoning
Requested-length error · UGI% off the requested word count, lower is better 156 · 91 of 230 measured 27.0% 18.0%no reasoning
Style adherence · UGIscore from 0 to 1, higher is better 24 · 108 of 230 measured 0.39 0.35no reasoning
Show all 22 results (13 more)
MeasureMistral Medium 3.1Mistral Medium 3.5
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results
Creative writing · TextArena rating, higher is better 123 · 95 of 385 measured 1,378 1,400
Expert prompts · TextArena rating, higher is better 143 · 119 of 338 measured 1,410 1,435
Instruction following · TextArena rating, higher is better 133 · 92 of 387 measured 1,395 1,422
Writing, literature and language · TextArena rating, higher is better 128 · 100 of 386 measured 1,386 1,405
Artificial Analysis reported by Artificial Analysis · 9 results
Terminal-Bench 2.1% of tasks, higher is better 97 · 63 of 131 measured 13.9% 50.6%
Terminal-Bench 4.0% of tasks, higher is better 72 · 72 of 120 measured 0.0% 0.0%
τ-bench banking% of tasks, higher is better 86 · 60 of 121 measured 8.2% 15.1%
Artificial Analysis Coding Indexindex score, higher is better 101 · 61 of 130 measured 20.5 46.9
IFBench% of instructions, higher is better 135 · 59 of 195 measured 39.8% 68.8%
Long-context reasoning (AA-LCR)% of questions, higher is better 176 · 120 of 241 measured 42.7% 69.3%
SciCode% of problems, higher is better 104 · 85 of 121 measured 32.4% 40.2%
Terminal-Bench Hard% of tasks, higher is better 127 · 57 of 190 measured 10.6% 33.3%
τ²-bench telecom% of tasks, higher is better 116 · 25 of 193 measured 40.6% 94.2%

Quick answers

Is Mistral Medium 3.1 better than Mistral Medium 3.5?

Mistral Medium 3.5 is the newer version and scores higher (Intelligence Index 14.2 against 9.2), but Mistral Medium 3.1 is cheaper (3.8 times less per token): upgrade if the gain is worth the price.

Which is cheaper, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.1 is cheaper: $0.80 against $3.00 per million tokens, 3.8 times less.

Which is better for agents and tool use, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.5 ranks 100 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 128 for Mistral Medium 3.1.

Which is better for reasoning and knowledge, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.5 ranks 123 of 187 models on sale for hard questions across science, maths and general knowledge, against 151 for Mistral Medium 3.1.

Which is better for writing, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.5 ranks 85 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 110 for Mistral Medium 3.1.

Which is better for sticking to the facts, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.5 ranks 98 of 139 models on sale for summarising without inventing things, and factual answers, against 122 for Mistral Medium 3.1.

Which has the bigger context window, Mistral Medium 3.1 or Mistral Medium 3.5?

Mistral Medium 3.5: 262,144 tokens, against 131,072 for Mistral Medium 3.1.

Run Mistral Medium 3.1 and Mistral Medium 3.5 on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.