The arena · AI model review
Mistral Large 3
by Mistral AI
Open-weight 675B MoE flagship (41B active) with 256K context, vision input, and $0.50/$1.50 per 1M pricing.
$1.50/1M out
Single 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.
Mistral AI
256K tokens
$0.50/1M in
$1.50/1M out
text, vision (image input), text output
Yes
What is Mistral Large 3?
Mistral AI's open-weight flagship released December 2, 2025: a sparse mixture-of-experts model with 675B total and 41B active parameters, 256K context, and text plus image input under Apache 2.0. Priced at $0.50/$1.50 per 1M tokens on La Plateforme, it debuted #2 among open-source non-reasoning models on LMArena.
Mistral Large 3 pros & cons
Pros
- 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
Cons
- 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
The arena’s verdict
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.
Thumbs up or thumbs down
Cast your verdict
Would you recommend Mistral Large 3, or warn the crowd away?
Top Mistral Large 3 alternatives
All alternativesAnthropic's fastest model: about 90% of Sonnet 4.5's coding skill at $1/$5 per 1M tokens, 200K context.
Anthropic's April 2026 Opus: 87.6% SWE-bench Verified, 1M context, high-res vision, now behind Opus 4.8
Anthropic's flagship Opus-tier model for long-horizon agentic coding; 1M context at $5/$25 per 1M tokens.
Compare Mistral Large 3 head-to-head
Mistral Large 3: frequently asked questions
How much does Mistral Large 3 cost per 1M tokens?
Mistral Large 3 costs $0.50/1M in per 1M input tokens and $1.50/1M out per 1M output tokens. Single 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.