Skip to content
llmwise
Sign inStart free

Mistral Large 4 · Writing

Mistral Large 4 for writing: our test runs

Mistral Large 4 ran our 5 writing prompts in llmwise, through the same pipeline your messages take. Here's every reply as it came, whether it passed, what it cost and what it counts as on Pro, next to the other models.

Start freeSee pricing

Based on 5 of our test runs on , through OpenRouter with the app's own prompt and settings. Updated .

Short answer

In our test runs on October 9, 2026, Mistral Large 4 passed 4 of 5 of our writing prompts (1 of the 2 hard ones): tied 4th of 19 models. Each reply counted as 1 message on Pro, where it has 250 a month, and cost $0.0024 to run on average.

Prompt by prompt

Each of the 5 writing prompts, and how Mistral Large 4's reply did.

Mistral Large 4 on each writing prompt
PromptResultCostTime
Announce a second bakery shop on LinkedInEverydayPassed: Graded 4.3 of 5 on average (lowest 3).$0.00148.9 s
Rewrite corporate jargon in plain wordsEverydayPassed: Graded 4.3 of 5 on average (lowest 4).$0.003034.3 s
Decline a meeting and offer two timesEverydayPassed: Graded 5.0 of 5 on average (lowest 5).$0.00093.9 s
A product announcement with five rulesHardFailed: Graded 3.3 of 5 on average (lowest 2); but doesn't end with a question.$0.004941.6 s
Argue both sides of free busesHardPassed: Graded 4.0 of 5 on average (lowest 4).$0.00189.8 s

Every writing reply from Mistral Large 4

  1. Announce a second bakery shop on LinkedIn

    An everyday prompt: “Announce a second bakery shop on LinkedIn”, in full.

    Passed: Graded 4.3 of 5 on average (lowest 3).

    Something warm is coming to Leeds. On 14 November, Crumb & Co. opens the doors of our second bakery, and we could not be happier to finally be part of your neighbourhood.

We started Crumb & Co. with a simple belief: good bread, made properly, brings people together. Seeing that belief grow from one small shop into a second home means everything to us — and to the wonderful team who bake, serve, and smile every single morning.

To say thank you for all the support, we're offering free coffee to everyone who visits us on opening day. Come say hello, meet the team, and help us settle in.

14 November. We'll save you a seat by the window.

    397 tokens in, 557 out (433 of them reasoning) · 8.9 s · $0.0014 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

  2. Rewrite corporate jargon in plain words

    An everyday prompt: “Rewrite corporate jargon in plain words”, in full.

    Passed: Graded 4.3 of 5 on average (lowest 4).

    During Q3, our teams collaborated across departments to focus on customers. This shift improved customer retention by 12% and cut churn by 0.3 percentage points among business clients. We also fine-tuned our customer service channels across all platforms to use resources more efficiently.

    427 tokens in, 1,273 out (1,369 of them reasoning) · 34.3 s · $0.0030 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

  3. Decline a meeting and offer two times

    An everyday prompt: “Decline a meeting and offer two times”, in full.

    Passed: Graded 5.0 of 5 on average (lowest 5).

    Subject: Re: Vendor Demo – Wednesday at 11:00

Hi Tom,

Thanks for the invitation to the vendor demo. Unfortunately, I can't make Wednesday at 11:00, as I'm presenting to the board at that time.

Would Thursday at 10:00 or Friday at 14:00 work for you instead? Happy to reschedule around either.

Best regards,
[Your name]

    400 tokens in, 295 out (219 of them reasoning) · 3.9 s · $0.0009 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

  4. A product announcement with five rules

    A hard prompt: “A product announcement with five rules”, in full.

    Failed: Graded 3.3 of 5 on average (lowest 2); but doesn't end with a question.

    Introducing the Nordvik 2 desk lamp, designed for precise control over your workspace lighting. The lamp dims smoothly across five distinct steps, allowing you to set the perfect brightness for any task. It charges from USB-C, keeping your desk clutter-free while delivering reliable power. The adjustable arm rotates 270 degrees, giving you complete flexibility to direct light exactly where you need it. Available for €79, the Nordvik 2 combines elegant design with practical functionality. Orders placed now will ship on 1 December, arriving in time for the winter season. Would you like to be among the first to experience the Nordvik 2 in your home or office.

    412 tokens in, 2,208 out (1,845 of them reasoning) · 41.6 s · $0.0049 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

  5. Argue both sides of free buses

    A hard prompt: “Argue both sides of free buses”, in full.

    Passed: Graded 4.0 of 5 on average (lowest 4).

    Free buses remove fares, speed boarding, cut congestion and emissions, help low-income residents, boost the local economy, and treat transit as a public good like roads and sidewalks. Every rider saves money, more people choose buses over cars, streets get safer, and cities can recoup costs through fuel taxes, parking fees, and productivity gains.

Free buses are not free; they shift costs to taxpayers, often through regressive taxes, while fares help fund maintenance and expansion. Zero fares can attract disruptive riders, raise security costs, and reduce service quality as demand outpaces capacity. Wealthy riders benefit most from subsidies, while drivers and businesses still pay for roads. Better to target discounts to those who need them and keep modest fares to preserve reliability, safety, and fiscal discipline.

    391 tokens in, 751 out (650 of them reasoning) · 9.8 s · $0.0018 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

How it compares on writing

Mistral Large 4 beside the job's picks among every model. 3 models share the hard-problems pick, and Mistral Large 4 isn't one of them.

Mistral Large 4 beside other models on writing
ModelPassedOn ProCost per reply
DeepSeek V4 ProOur pick: hard problems (shared), best value5 of 5250 a month on Pro$0.0011
GPT-6 LunaOur pick: hard problems (shared), everyday5 of 560 a day on Pro$0.0001
Mistral Large 44 of 5250 a month on Pro$0.0024
On Pro: Pro's count on each model. Cost: what OpenRouter charged us per reply, on average.

Every limit is published. Paid plans also have a monthly fair-use limit on AI cost: Pro $7.50, Max $20, Ultra $42, Studio $85. Using every message on your plan at typical sizes stays under it; very large messages and heavy research use it faster. Every limit, explained.

Mistral Large 4 next to each model on writing

Mistral Large 4 against each model on the same 5 writing prompts. One run each, through OpenRouter: a wait depends on the provider and the load that day, so a lead under 10% counts as close.

  • Against Claude Fable 5.1: Mistral Large 4 answered 1.6× later (9.8 s to 6.0 s), cost 7.2× less, and passed 4 of 5 to its 3. Only Mistral Large 4 passed Argue both sides of free buses.

  • Against Claude Opus 5.5: Mistral Large 4 answered 1.2× later (9.8 s to 8.2 s), cost 7.7× less, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Argue both sides of free buses. Only Claude Opus 5.5 passed A product announcement with five rules.

  • Against Claude Sonnet 5.5: Mistral Large 4 answered 4.2× later (9.8 s to 2.3 s), cost 1.4× less, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Argue both sides of free buses. Only Claude Sonnet 5.5 passed A product announcement with five rules.

  • Against Claude Sonnet 5: Mistral Large 4 answered 2.2× later (9.8 s to 4.5 s), cost 1.4× less, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Announce a second bakery shop on LinkedIn. Only Claude Sonnet 5 passed A product announcement with five rules.

  • Against Claude Haiku 5.5: Mistral Large 4 answered 6.4× later (9.8 s to 1.5 s), cost 14.9× more, and passed 4 of 5 to its 3. Only Mistral Large 4 passed Rewrite corporate jargon in plain words; Argue both sides of free buses. Only Claude Haiku 5.5 passed A product announcement with five rules.

  • Against Claude Haiku 4.5: Mistral Large 4 answered 4.4× later (9.8 s to 2.2 s), cost 2.1× more, and passed 4 of 5 to its 4.

  • Against GPT-6 Astra: Mistral Large 4 answered 2.1× later (9.8 s to 4.8 s), cost 4.6× less, and passed 4 of 5 to its 5. Only GPT-6 Astra passed A product announcement with five rules.

  • Against GPT-6.1 Sol: Mistral Large 4 answered 3.0× later (9.8 s to 3.3 s), cost 2.1× more, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Rewrite corporate jargon in plain words. Only GPT-6.1 Sol passed A product announcement with five rules.

  • Against GPT-6 Sol: Mistral Large 4 answered 2.8× later (9.8 s to 3.5 s), cost within 10% of it, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Rewrite corporate jargon in plain words. Only GPT-6 Sol passed A product announcement with five rules.

  • Against GPT-6 Luna: Mistral Large 4 answered 4.1× later (9.8 s to 2.4 s), cost 24.1× more, and passed 4 of 5 to its 5. Only GPT-6 Luna passed A product announcement with five rules.

  • Against Gemini 3.1 Pro (preview): Mistral Large 4 answered within 10% of its time (9.8 s to 9.1 s), cost 3.8× less, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Argue both sides of free buses. Only Gemini 3.1 Pro (preview) passed A product announcement with five rules.

  • Against Gemini 3.8 Flash: Mistral Large 4 answered 2.5× later (9.8 s to 3.9 s), cost 1.8× more, and passed 4 of 5 to its 3. Only Mistral Large 4 passed Rewrite corporate jargon in plain words.

  • Against DeepSeek V4.1 Flash: Mistral Large 4 answered 7.4× later (9.8 s to 1.3 s), cost 5.3× more, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Rewrite corporate jargon in plain words. Only DeepSeek V4.1 Flash passed A product announcement with five rules.

  • Against DeepSeek V4 Pro: Mistral Large 4 answered 2.5× later (9.8 s to 3.9 s), cost 2.2× more, and passed 4 of 5 to its 5. Only DeepSeek V4 Pro passed A product announcement with five rules.

  • Against Grok 4.7: Mistral Large 4 answered 1.6× later (9.8 s to 6.0 s), cost 2.4× less, and passed 4 of 5 to its 3. Only Mistral Large 4 passed Rewrite corporate jargon in plain words; Argue both sides of free buses. Only Grok 4.7 passed A product announcement with five rules.

  • Against Kimi K3: Mistral Large 4 answered 3.3× later (9.8 s to 3.0 s), cost 1.6× less, and passed 4 of 5 to its 4.

  • Against GLM 5.3: Mistral Large 4 answered 5.6× later (9.8 s to 1.8 s), cost 4.2× more, and passed 4 of 5 to its 4. Only Mistral Large 4 passed Argue both sides of free buses. Only GLM 5.3 passed A product announcement with five rules.

  • Against GLM 5.3 Flash: Mistral Large 4 answered 1.2× later (9.8 s to 8.5 s), cost 7.5× more, and passed 4 of 5 to its 2. Only Mistral Large 4 passed Rewrite corporate jargon in plain words; Argue both sides of free buses.

How these runs were done

Rubric (graded). The grader model scores the reply from 1 to 5 on each published criterion. It passes with an average of 4 or more and no criterion under 3, and only if it also meets the prompt's automatic rules (length, words it must or mustn't use).

The grader is Claude Opus 5.5 at low reasoning effort; its own replies are graded by GPT-6 Astra, so no model grades itself. Its prompt and every rubric are on the methods page.

How the runs were done, and every writing prompt.

Mistral Large 4, and writing, elsewhere

Questions

Is Mistral Large 4 good for writing?

In our test runs it passed 4 of 5 writing prompts, tied 4th of the 19 models in llmwise. Every reply is on this page, so you can judge them yourself.

How many of my messages does a writing reply from Mistral Large 4 use?

1 message each on Pro, where it has 250 a month on Pro. The price of a message is fixed and shown before you send it, however long the reply.

How were these runs done?

The same way for every model: each prompt sent through llmwise's own pipeline, each reply checked the same way. The methods page has every prompt and how each is scored.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, GLM, and Mistral, in one chat.

See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.

Start freeSee pricing