Mistral AI is a Paris-based lab building efficient, Apache 2.0-licensed open-weight models. Mistral Large 2 is preferred for compliance-sensitive European deployments. Codestral 25.01 is the strongest open-weight model for code generation.
Rankings refresh dailyScored on 6 criteriaNo paid rankings
Apache 2.0 license — use commercially with no restrictions
Preferred by European enterprises for data residency compliance
Codestral 25.01 is the strongest open-weight code model in 2026
24 models
All Mistral AI Models
Every Mistral AI model in the directory, ranked by overall capability score.
MistralBudget
Mistral: Mistral Medium 3.1
Mistral Medium 3.1 is a multimodal mid-tier model from Mistral that supersedes Mistral Large 2, offering vision capabilities alongside strong text performance at a significantly reduced price point. It targets the sweet spot between budget models and expensive flagships, with a 128K context window and competitive multilingual support.
Verdict
The best Mistral model for budget-conscious builders who still need multimodal capability and solid multilingual output.
Quality score
70%
Pricing
Mistral AI API Pricing
Per 1 million tokens. Updated when providers change prices.
Pricing changes, new releases, and ranking shifts — straight to your inbox.
No spam. Useful updates only. Affiliate disclosures always clearly labeled.
Mistral AI FAQ
What is Mistral's best model in 2026?
Mistral Large 2 is Mistral's most capable general model. Codestral 25.01 is Mistral's strongest coding model — competitive with Claude Sonnet 4.6 on specific coding benchmarks. Mistral Small 3.1 is an excellent free option for lightweight tasks.
Why choose Mistral over OpenAI or Anthropic?
Mistral models are Apache 2.0 licensed — you can self-host, fine-tune, and deploy commercially with no usage restrictions. This matters for European data residency, compliance requirements, and teams that need full model ownership.
Are Mistral models open source?
Yes — Mistral Small 3.1, Mistral Large 2, and Codestral are available under Apache 2.0. You can download, modify, and deploy them commercially without needing to pay or seek permission.
Best for cost-sensitive teams needing solid coding, instruction-following, and basic vision tasks without paying flagship prices.
Context
131k tokens
Officially supersedes Mistral Large 2, representing a generational shift in Mistral's lineup toward multimodal capability at lower cost tiers. Available via Mistral API and select cloud providers. No function calling limitations noted at this tier.
BudgetMultimodalMultilingualMid-tierVision
Best for
Cost-sensitive teams needing solid coding, instruction-following, and basic vision tasks without paying flagship prices.
Pixtral Large 2411 is Mistral's flagship multimodal model, adding native image understanding to the Mistral Large 2 foundation. It processes both text and images with strong reasoning across documents, charts, and visual content.
Verdict
A capable and fairly priced multimodal flagship, best suited for Mistral ecosystem users and European compliance requirements.
Quality score
74%
Pricing
$2.00/1M in
$6.00/1M out
Speed
Balanced
Best for teams needing a capable european-hosted multimodal model for document analysis, visual qa, and code generation with image context.
Context
131k tokens
Available via Mistral API (la Plateforme) and supports self-hosted deployment. The '2411' suffix indicates a November 2024 release. Supersedes Mistral Large 2 as the primary flagship. Image input pricing follows the same $2/1M token rate.
Mistral Small 4 is a compact, cost-efficient language model from Mistral AI that punches well above its price class, succeeding Mistral Large 2 in capability while costing a fraction of the price. It features a 256K context window and is optimized for high-throughput, latency-sensitive applications.
Verdict
The best bang-for-buck text model in its class — Mistral Large 2 quality at a fraction of the cost.
Quality score
68%
Pricing
$0.15/1M in
$0.60/1M out
Speed
Fast
Best for teams needing reliable, fast text generation and coding assistance at near-commodity pricing without sacrificing too much quality.
Context
262k tokens
Pricing at $0.15/$0.60 per million tokens makes this one of the most affordable capable models on the market. Available via Mistral's La Plateforme API and compatible with OpenAI-style endpoints. No image input support confirmed at launch.
BudgetFastLong ContextMultilingualCoding
Best for
Teams needing reliable, fast text generation and coding assistance at near-commodity pricing without sacrificing too much quality.
Mistral Large 3 2512 is Mistral's flagship dense model updated in December 2025, offering strong multilingual reasoning and coding capabilities at a significantly reduced price point compared to its predecessor. It targets enterprise workloads that need high-quality outputs without paying top-tier frontier model prices.
Verdict
The best price-per-quality ratio in the non-mini flagship tier, especially for multilingual and long-context enterprise tasks.
Quality score
69%
Pricing
$0.50/1M in
$1.50/1M out
Speed
Balanced
Best for multilingual enterprise tasks, code generation, and long-document analysis where cost efficiency matters more than absolute state-of-the-art performance.
Context
262k tokens
Pricing of $0.50 input / $1.50 output per 1M tokens places it firmly in the budget-flagship category. Available via Mistral API (La Plateforme) and major cloud providers. December 2025 update ('2512') improves instruction following over the earlier 2407 release.
Multilingual enterprise tasks, code generation, and long-document analysis where cost efficiency matters more than absolute state-of-the-art performance.
Mistral Small 3.2 24B is a compact 24-billion parameter model from Mistral that punches well above its weight class, superseding Mistral Large 2 at a fraction of the cost. It handles coding, instruction-following, and multilingual tasks with strong efficiency for its size.
Verdict
The best budget coding model available today, offering frontier-adjacent performance at commodity pricing.
Quality score
68%
Pricing
$0.07/1M in
$0.20/1M out
Speed
Fast
Best for high-volume production workloads where cost matters but quality can't be sacrificed entirely — especially code generation and structured output tasks.
Context
128k tokens
Mistral Small 3.2 is available as an open-weight model, making it deployable on-premises or via self-hosted infrastructure — a key differentiator over GPT-4o Mini and Claude Haiku for privacy-sensitive use cases.
BudgetCodingEfficientOpen-weightMultilingual
Best for
High-volume production workloads where cost matters but quality can't be sacrificed entirely — especially code generation and structured output tasks.
Mistral's ultra-budget multimodal model — exceptionally cheap with vision support, built for high-volume lightweight tasks where cost is the primary constraint.
Verdict
Ultra-cheap multimodal model for massive-volume, low-complexity pipelines.
Quality score
57%
Pricing
$0.03/1M in
$0.11/1M out
Speed
Very fast
Best for ultra-high-volume classification, summarisation, and lightweight vision tasks
Context
128k tokens
At $0.10/1M input, the cost question disappears. The only question is whether the task complexity exceeds what Mistral Small can handle.
BudgetMultimodalUltra cheapMistral
Best for
Ultra-high-volume classification, summarisation, and lightweight vision tasks
Mistral Medium 3 is a mid-tier model from Mistral AI that punches above its weight class, officially superseding Mistral Large 2 while costing a fraction of the price. It targets teams needing capable multilingual and coding performance without flagship-level spend.
Verdict
The most capable budget model Mistral has shipped — a smart default for high-volume teams that need real performance without flagship pricing.
Quality score
67%
Pricing
$0.40/1M in
$2.00/1M out
Speed
Fast
Best for cost-conscious teams running high-volume coding, summarization, or multilingual tasks at enterprise scale.
Context
131k tokens
Priced at $0.40 input / $2.00 output per 1M tokens. Officially supersedes Mistral Large 2, making it an easy drop-in upgrade for existing Mistral users. Available via Mistral's API and La Plateforme.
BudgetMultilingualCodingHigh VolumeMid-Tier
Best for
Cost-conscious teams running high-volume coding, summarization, or multilingual tasks at enterprise scale.
Balanced enterprise model with consistent reasoning, good speed, and a dependable middle-ground — especially for European teams with data residency requirements.
Verdict
Best balanced generalist for EU teams with data residency needs.
Quality score
67%
Pricing
$2.00/1M in
$6.00/1M out
Speed
Balanced
Best for balanced team usage with eu data residency requirements
Context
128k tokens
The EU hosting angle is the real differentiator here — for teams outside Europe, other models perform better.
EU hostingBalancedTeam default
Best for
Balanced team usage with EU data residency requirements
Devstral Medium is Mistral's code-focused model optimized for software development tasks, offering strong code generation and debugging capabilities at a budget-friendly price point. It targets developers who need reliable coding assistance without paying flagship model rates.
Verdict
A genuinely specialized, budget-friendly coding model that earns its place in any developer's API toolkit.
Quality score
60%
Pricing
$0.40/1M in
$2.00/1M out
Speed
Balanced
Best for developers seeking capable code generation, debugging, and code review at a fraction of the cost of gpt-4-class models.
Context
131k tokens
Pricing is notably aggressive at ~$0.40 input / $2.00 output per 1M tokens. Available via Mistral's La Plateforme API. Part of the Devstral family, which is distinct from Mistral's general-purpose Mistral Medium line.
Mistral Nemo is a compact 12B-parameter open-weight model developed in collaboration with NVIDIA, designed to deliver strong multilingual and instruction-following performance at an extremely low cost. It fits into a 128K context window and is optimized for deployment efficiency without sacrificing too much reasoning depth.
Verdict
A dirt-cheap multilingual model perfect for bulk text tasks, but don't expect frontier-level reasoning.
Quality score
55%
Pricing
$0.02/1M in
$0.03/1M out
Speed
Fast
Best for teams needing a cheap, fast, multilingual workhorse for classification, summarization, or light coding tasks at scale.
Context
131k tokens
Mistral Nemo is open-weight (Apache 2.0 license), so self-hosting is an option for teams that want to eliminate API costs entirely. Pricing via API is through Mistral's La Plateforme. The model uses a Tekken tokenizer which is more efficient than older Mistral tokenizers, especially for non-English text.
budgetmultilingualopen-weight12Befficient
Best for
Teams needing a cheap, fast, multilingual workhorse for classification, summarization, or light coding tasks at scale.
Devstral 2 2512 is Mistral's second-generation code-specialized model, built specifically for software development tasks with a 256K context window. It targets developers needing a cost-efficient coding assistant without sacrificing meaningful capability.
Verdict
A purpose-built coding workhorse that punches well above its price tag for development teams running high-volume or agentic pipelines.
Quality score
55%
Pricing
$0.40/1M in
$2.00/1M out
Speed
Fast
Best for budget-conscious developers who need a capable coding model for agentic workflows, code generation, and repository-scale context at a fraction of flagship pricing.
Context
262k tokens
The December 2025 (2512) release date suggests this is a recent iteration. Pricing at $0.40 input / $2.00 output is notably competitive for a code-specialist model with 256K context. Verify availability and rate limits via Mistral API or partner providers.
Code-specialistBudgetLong contextAgenticMistral
Best for
Budget-conscious developers who need a capable coding model for agentic workflows, code generation, and repository-scale context at a fraction of flagship pricing.
Devstral Small 1.1 is Mistral's code-specialized small model, purpose-built for software engineering tasks including code generation, debugging, and repository-level reasoning. It succeeds Devstral Small 1.0 with improved instruction following and agentic coding capabilities at a fraction of flagship model costs.
Verdict
The best dollar-for-dollar coding model for agentic pipelines that doesn't need to do anything else.
Quality score
54%
Pricing
$0.10/1M in
$0.30/1M out
Speed
Fast
Best for developers who need a cheap, fast coding assistant for agentic workflows, code review, and multi-file repo tasks without paying flagship prices.
Context
131k tokens
Available via Mistral API and can be self-hosted via open weights. Pricing is among the lowest available for a code-specialized model. Designed to work within coding agent frameworks like SWE-agent and OpenHands.
Codestral 2508 is Mistral's latest dedicated code model, succeeding Codestral 25.01 with improved code generation, completion, and reasoning across 80+ programming languages. It offers a massive 256K context window at a budget-friendly price point aimed squarely at developer tooling and IDE integrations.
Verdict
The most cost-effective specialized code model for production developer tooling with serious context capacity.
Quality score
53%
Pricing
$0.30/1M in
$0.90/1M out
Speed
Fast
Best for high-volume code generation, completion, and refactoring tasks where cost efficiency and long-context handling matter most.
Context
256k tokens
Available via Mistral's La Plateforme API. Also accessible through Continue.dev, Cursor, and other IDE integrations that support the Codestral endpoint. FIM (fill-in-the-middle) mode is specifically supported for autocomplete use cases. Output price rounds to ~$0.90/1M tokens.
Ministral 3B is Mistral's ultra-compact edge model designed for low-latency, cost-sensitive deployments. It punches above its weight for a sub-4B parameter model, handling instruction following, summarization, and lightweight reasoning at near-negligible cost.
Verdict
The go-to model for bulk processing tasks where cost and speed trump quality.
Quality score
50%
Pricing
$0.15/1M in
$0.15/1M out
Speed
Very fast
Best for high-volume, latency-sensitive applications where cost per token matters more than top-tier quality.
Context
262k tokens
The '8B 2512' in the model name likely refers to a specific versioned release; despite the naming, this is based on Mistral's 3B architecture. Confirm parameter count and capabilities with Mistral's official documentation before production use.
budgetedgefastlong-contextcompact
Best for
High-volume, latency-sensitive applications where cost per token matters more than top-tier quality.
Mistral Small 3 is a compact, budget-oriented language model from Mistral AI that punches above its weight class for everyday NLP tasks. It supersedes Mistral Large 2 in efficiency while targeting cost-sensitive deployments that don't require frontier-level reasoning.
Verdict
A lean, fast, affordable workhorse for text tasks — ideal for scale, not for depth.
Quality score
55%
Pricing
$0.05/1M in
$0.08/1M out
Speed
Very fast
Best for high-volume, cost-sensitive applications like customer support automation, content drafting, and lightweight code assistance.
Context
33k tokens
Pricing is exceptionally competitive at $0.05/$0.08 per 1M tokens. Available via Mistral's La Plateforme API and various third-party providers. GDPR-friendly EU-based hosting is a notable advantage for European enterprise customers. No image input or output support.
BudgetFastMultilingualLightweightHigh-volume
Best for
High-volume, cost-sensitive applications like customer support automation, content drafting, and lightweight code assistance.
Ministral 3B is Mistral's compact edge-optimized model designed for high-throughput, low-latency tasks at an extremely competitive price point. Despite its small size, it supports a 262K context window, making it unusually capable for a sub-$0.20/1M token model.
Verdict
An ultra-cheap, fast model with a surprisingly large context window, but quality limitations make it a pipeline tool rather than a general assistant.
Quality score
48%
Pricing
$0.20/1M in
$0.20/1M out
Speed
Very fast
Best for high-volume, cost-sensitive workflows like document triage, classification, summarization, and lightweight coding assistance where budget is the primary constraint.
Context
262k tokens
Model name suggests a December 2025 revision ('2512'). Pricing is symmetric at $0.20/1M for both input and output, which simplifies cost modeling. Confirm availability on your target API platform as Mistral model availability varies by provider.
budgetedgesmall modellong contexthigh throughput
Best for
High-volume, cost-sensitive workflows like document triage, classification, summarization, and lightweight coding assistance where budget is the primary constraint.
Mixtral 8x22B Instruct is Mistral's flagship sparse mixture-of-experts model, routing tokens through 2 of 8 expert networks (39B active parameters out of 141B total) for efficient high-quality inference. It excels at multilingual tasks, code generation, and instruction-following with strong European language support.
Verdict
A capable MoE workhorse with strong multilingual chops, but its short context window and rising competition have eroded its value proposition.
Quality score
59%
Pricing
$2.00/1M in
$6.00/1M out
Speed
Balanced
Best for teams needing strong multilingual capabilities and solid coding performance at a mid-tier price point without relying on openai or anthropic infrastructure.
Context
66k tokens
Available via Mistral API and as open weights (Apache 2.0 license) for self-hosting. The open-weight option is a key differentiator for privacy-sensitive or on-premise deployments. API pricing at $2/$6 per million tokens is mid-range but faces pressure from newer, cheaper alternatives.
MoEMultilingualOpen-weightMid-tierInstruct
Best for
Teams needing strong multilingual capabilities and solid coding performance at a mid-tier price point without relying on OpenAI or Anthropic infrastructure.
Voxtral Small 24B is Mistral's audio-capable language model, designed for speech transcription, voice understanding, and spoken language tasks at a budget-friendly price point. It supersedes Mistral Small 3.1 with native audio input support built on a 24B parameter base.
Verdict
A purpose-built budget audio model that excels at voice tasks but stumbles on context length and general-purpose depth.
Quality score
47%
Pricing
$0.10/1M in
$0.30/1M out
Speed
Fast
Best for transcribing, analyzing, and responding to audio input cost-effectively without needing a separate speech-to-text pipeline.
Context
32k tokens
Voxtral Small is audio-in capable but does not support image input. The 32K context window is notably short for a 2025 model. Pricing is via Mistral's API; availability through third-party providers may vary. Check whether your use case requires audio input — the text-only version of Mistral Small 3.1 may be more appropriate for pure text workloads.
Audio AIBudgetMultilingualSpeechMistral
Best for
Transcribing, analyzing, and responding to audio input cost-effectively without needing a separate speech-to-text pipeline.
Mixtral 8x7B Instruct is Mistral's sparse mixture-of-experts model that routes tokens through 2 of 8 expert networks, achieving strong performance while activating only ~13B parameters per forward pass. It excels at instruction-following, multilingual tasks, and code generation at a competitive price point.
Verdict
A historically significant open-weight model that's been surpassed by newer alternatives but still earns its place in self-hosted and multilingual pipelines.
Quality score
53%
Pricing
$0.54/1M in
$0.54/1M out
Speed
Fast
Best for developers and teams needing a capable open-weight model for coding, multilingual tasks, and general instruction-following without flagship model pricing.
Context
33k tokens
Pricing is symmetric at $0.54/1M for both input and output. As an open-weight model, costs can drop significantly if self-hosted. The 32K context window is a hard ceiling — plan accordingly for document-heavy workflows.
Developers and teams needing a capable open-weight model for coding, multilingual tasks, and general instruction-following without flagship model pricing.
Ministral 3B is Mistral's ultra-compact 3-billion parameter edge model designed for lightweight inference, on-device deployment, and cost-sensitive applications. It delivers surprisingly capable text understanding and generation at a fraction of the cost of larger models.
Verdict
The cheapest viable option for simple NLP tasks, but don't expect small-flagship performance.
Quality score
41%
Pricing
$0.10/1M in
$0.10/1M out
Speed
Very fast
Best for high-volume, low-latency tasks where cost and speed matter more than frontier-level reasoning.
Context
131k tokens
Priced at a flat $0.10/1M for both input and output, making cost estimation predictable. The '2512' suffix indicates a December 2025 release version. Best suited for batch processing, classification, or extraction pipelines where volume is high and task complexity is low.
3BEdgeUltra-budgetMistralLightweight
Best for
High-volume, low-latency tasks where cost and speed matter more than frontier-level reasoning.
Mistral Saba is a compact, budget-oriented language model from Mistral designed for efficient text tasks with a focus on Arabic and South Asian languages alongside English. It targets cost-sensitive deployments where multilingual support is more important than raw reasoning depth.
Verdict
A bargain multilingual model built for Arabic and South Asian languages, but too constrained for demanding workloads.
Quality score
45%
Pricing
$0.20/1M in
$0.60/1M out
Speed
Fast
Best for low-cost multilingual applications requiring arabic, hindi, or urdu language support
Context
33k tokens
Pricing reflects Mistral API rates and may vary by reseller. The model's name 'Saba' references Arabic linguistic heritage, signaling its intended multilingual focus. No vision or tool-use capabilities documented at launch.
BudgetMultilingualArabicCompactEfficient
Best for
Low-cost multilingual applications requiring Arabic, Hindi, or Urdu language support
Mistral Small Creative is a fine-tuned variant of Mistral Small optimized for creative writing tasks, offering a budget-friendly option for generative content at under $0.10/1M input tokens. It targets storytelling, copywriting, and imaginative text generation at a fraction of the cost of flagship models.
Verdict
A lean, cheap creative writing workhorse — ideal for volume content generation but not for quality-critical storytelling.
Quality score
36%
Pricing
$0.10/1M in
$0.30/1M out
Speed
Fast
Best for budget-conscious creative writing tasks like short stories, marketing copy, and brainstorming where cost matters more than peak quality.
Context
33k tokens
Context window of 32,768 tokens is notably smaller than competing budget models. Pricing is approximate ($0.10 input / $0.30 output per 1M tokens). Availability is through Mistral's API (La Plateforme) and may also be accessible via third-party providers. Confirm fine-tune scope before deploying for non-creative tasks.
Creative WritingBudgetFastShort-formMistral
Best for
Budget-conscious creative writing tasks like short stories, marketing copy, and brainstorming where cost matters more than peak quality.
Mistral 7B Instruct v0.1 is a 7-billion-parameter instruction-tuned model from Mistral AI, one of the earliest open-weight models to challenge larger proprietary models on efficiency. It handles general text tasks at extremely low cost but is constrained by a very small context window of under 3K tokens.
Verdict
A historically significant but now outdated budget model crippled by an unusably small context window.
Quality score
26%
Pricing
$0.11/1M in
$0.19/1M out
Speed
Very fast
Best for ultra-low-cost simple text tasks like classification, short summarization, or lightweight chatbot responses where context length is not a concern.
Context
3k tokens
This is v0.1, the original release — not to be confused with v0.2 or v0.3 which substantially improve context length and quality. The listed context window of ~2,824 tokens is unusually small even among budget models. Marked as superseding Mistral Large 2 in the spec, which appears to be a data error — this model does not supersede Mistral Large 2 in capability or positioning.
budgetopen-weightsmall modellegacyfast
Best for
Ultra-low-cost simple text tasks like classification, short summarization, or lightweight chatbot responses where context length is not a concern.
