Head-to-head

Mistral Large 3 logovsMistral Large 4 logo

Mistral Large 3 vs Mistral Large 4: which AI model wins in 2026?

Mistral Large 3 ($1.50/1M out) and Mistral Large 4 ($4.18/1M out (list)) are two of the most-used AI models in 2026. Compare them line by line below, then cast your verdict.

Quick verdict

On Reasoning, pick Mistral Large 4: the arena rates it 4/5 against 3/5 for Mistral Large 3. On budget, Mistral Large 3 wins: it starts at $1.50/1M out versus $4.18/1M out (list) for Mistral Large 4.

Line-by-line comparison

From
$1.50/1M outSingle API tier on La Plateforme ($0.50 in / $1.50 out per 1M tokens), with a 50% discount via the batch API; Apache 2.0 weights allow free self-hosting, and third-party host pricing may differ.
$4.18/1M out (list)Public preview until ~Oct 20 at 50% off: $0.68 input / $2.09 output per 1M tokens. List price is $1.36 / $4.18. Cached input $0.07/1M during preview. Weights due end of Oct 2026 under open license. Verified against mistral.ai on 2026-10-13.
Provider
Mistral AI
Mistral AI
Context window
256K tokens
520K-1M tokens (varies by source)
Input price
$0.50/1M in
$1.36/1M in (list)
Output price
$1.50/1M out
$4.18/1M out (list)
Modalities
text, vision (image input), text output
text, vision (image input), text output
Open weights
Yes
Yes
Crowd score
50%(0)
50%(0)
Arena ratings (1-5)
Reasoning
3.0
4.0
Coding
3.0
3.5
Writing
4.0
4.0
Speed
3.5
3.0
Value
4.0
3.5

Strengths and weaknesses

Mistral Large 3

  • Apache 2.0 open weights with single-node deployment via FP8/NVFP4 quantization, despite 675B total parameters
  • 256K context window, at the upper end for open-weight models, well suited to long-document RAG
  • Aggressive flagship pricing at $0.50 in / $1.50 out per 1M tokens, roughly 3-4x cheaper than Western proprietary flagships
  • Debuted #2 among open-source non-reasoning models on LMArena (Elo ~1418)
  • Native multimodality (2.5B-parameter vision encoder) and 40+ native languages
  • Developers on HN praise its strict formatting and instruction following plus production reliability
  • Weak deep reasoning: GPQA Diamond ~44% vs high-70s for DeepSeek V3.2 and Kimi K2 Thinking; no reasoning variant at launch
  • Trails GLM-4.6, Kimi K2 and DeepSeek on modern coding benchmarks (middling LiveCodeBench v6); HN devs place it a 'different weight class' below Gemini 3, GPT-5.1 and Claude Opus 4.5
  • Hallucination-prone on factual QA (SimpleQA ~24%) with weak abstention tuning
  • Measured output speed ~49 tok/s on Artificial Analysis, below the ~58 tok/s median for comparable models
  • HN criticism that the architecture closely mirrors DeepSeek V3, raising doubts about original R&D

Mistral Large 4

  • First European trillion-parameter model, trained on EU infrastructure with full GDPR compliance
  • Open weights coming by end of Oct 2026, enabling self-hosting and fine-tuning under permissive license
  • 93% on Cybench and 82% on Artificial Analysis Cyber Index, the highest cybersecurity scores at launch
  • Natively multimodal with superior visual grounding (beats GPT-6-Astra on Dense 200 benchmark)
  • 59.9% on AutomationBench (657 business workflows), ahead of Kimi K3 and DeepSeek V4 Pro
  • 160+ native languages and aggressive preview pricing at $0.68/$2.09 (50% off list through mid-Oct)
  • Preview pricing ends around Oct 20; list price of $1.36/$4.18 is higher than Chinese competitors
  • Context window unclear: docs say 1M but Artificial Analysis and OpenRouter report 520K for preview
  • Language score 49.6/100 on LiveBench, lowest category, multilingual claims need independent validation
  • Weights not yet released; self-hosting benefit is promised, not delivered
  • Architecture heavily mirrors DeepSeek V3/V4, raising questions about original R&D contribution

Cast your verdict

One recommendation per tool per gladiator. It reshapes the crowd score everyone sees.

Mistral Large 3$1.50/1M out
50%crowd score · 0
Mistral Large 4$4.18/1M out (list)
50%crowd score · 0

The arena’s verdict on Mistral Large 3

Choose Mistral Large 3 if you want an open-weight, EU-governed flagship for multilingual RAG, long-document work, or self-hosted deployments: versus Mistral Large 2 (dense 123B, restrictive research license, $2/$6 API pricing) it is a clear upgrade in context, multimodality, licensing, and cost. Avoid it as your primary coding or deep-reasoning engine; DeepSeek V3.2, GLM-4.6, or proprietary frontier models score materially higher there. Treat it as a cheap, reliable workhorse rather than a frontier performer.

The arena’s verdict on Mistral Large 4

Pick Mistral Large 4 if you need a trillion-scale, EU-governed model for cybersecurity, multimodal automation, or if you are waiting for open weights to self-host. The preview pricing is aggressive, and the cybersecurity benchmarks are unmatched. Avoid it if you need peak coding or deep reasoning: DeepSeek V4 Pro and Claude Opus 5 score higher there. Wait for weights release (late Oct) and context-window clarification before committing to production workloads. Compare to Mistral Large 3 ($0.50/$1.50) if you do not need vision or the extra scale.

Frequently asked questions

Is Mistral Large 3 better than Mistral Large 4?

On Reasoning, Mistral Large 4 rates higher (4/5 vs 3/5). The right pick depends on your use case. The line-by-line comparison on this page breaks down pricing, key specs and arena ratings.

Which is cheaper, Mistral Large 3 or Mistral Large 4?

Mistral Large 3 is cheaper: it starts at $1.50/1M out, while Mistral Large 4 starts at $4.18/1M out (list).

How much do Mistral Large 3 and Mistral Large 4 cost per 1M tokens?

Mistral Large 3: $0.50/1M in per 1M input tokens, $1.50/1M out per 1M output tokens. Mistral Large 4: $1.36/1M in (list) per 1M input tokens, $4.18/1M out (list) per 1M output tokens.