LLM Reference

Mistral Large 2 vs Mistral Small 4

Mistral Large 2 (2025) and Mistral Small 4 (2026) are compact production models from MistralAI. Mistral Large 2 ships a 128K-token context window, while Mistral Small 4 ships a 256K-token context window. On pricing, Mistral Small 4 costs $0.15/1M input tokens versus $0.48/1M for the alternative. This comparison covers specs, pricing, capabilities, benchmarks, provider availability, and production fit. It focuses on practical selection signals rather than broad model-family marketing.

Mistral Small 4 is ~220% cheaper at $0.15/1M; pay for Mistral Large 2 only for vision-heavy evaluation.

Decision scorecard

Local evidence first
SignalMistral Large 2Mistral Small 4
Decision fitCoding, RAG, and AgentsRAG, Agents, and Long context
Context window128K256K
Cheapest output$2.4/1M tokens$0.6/1M tokens
Provider routes4 tracked3 tracked
Shared benchmarks0 rows0 rows

Decision tradeoffs

Choose Mistral Large 2 when...
  • Mistral Large 2 has broader tracked provider coverage for fallback and procurement flexibility.
  • Mistral Large 2 uniquely exposes Structured outputs in local model data.
  • Local decision data tags Mistral Large 2 for Coding, RAG, and Agents.
Choose Mistral Small 4 when...
  • Mistral Small 4 has the larger context window for long prompts, retrieval packs, or transcript analysis.
  • Mistral Small 4 has the lower cheapest tracked output price at $0.6/1M tokens.
  • Local decision data tags Mistral Small 4 for RAG, Agents, and Long context.

Monthly cost at traffic

Estimate token spend from the cheapest tracked input and output prices on this page.

Lower estimate Mistral Small 4

Mistral Large 2

$984

Cheapest tracked route: AWS Bedrock

Mistral Small 4

$270

Cheapest tracked route: OpenRouter

Estimated monthly gap: $714. Batch, cache, and negotiated pricing are excluded from this local estimate.

Switch friction

Mistral Large 2 -> Mistral Small 4
  • Provider overlap exists on OpenRouter and Mistral AI Studio; start route-level A/B tests there.
  • Mistral Small 4 is $1.8/1M tokens lower on cheapest tracked output pricing before cache, batch, or negotiated discounts.
  • Check replacement coverage for Structured outputs before moving production traffic.
Mistral Small 4 -> Mistral Large 2
  • Provider overlap exists on OpenRouter and Mistral AI Studio; start route-level A/B tests there.
  • Mistral Large 2 is $1.8/1M tokens higher on cheapest tracked output pricing, so quality gains need to justify the spend.
  • Mistral Large 2 adds Structured outputs in local capability data.

Specs

Specification
Released2025-11-252026-03-16
Context window128K256K
Parameters123B119B (6.5B active)
Architecturedecoder onlymoe
LicenseTrueApache 2.0
Knowledge cutoff2025-072025-06

Pricing and availability

Pricing attributeMistral Large 2Mistral Small 4
Input price$0.48/1M tokens$0.15/1M tokens
Output price$2.4/1M tokens$0.6/1M tokens
Providers

Capabilities

CapabilityMistral Large 2Mistral Small 4
VisionYesYes
MultimodalYesYes
ReasoningNoNo
Function callingYesYes
Tool useYesYes
Structured outputsYesNo
Code executionNoNo

Benchmarks

No shared benchmark rows are currently sourced for this pair.

Deep dive

The capability footprint differs most on structured outputs: Mistral Large 2. Both models share vision, multimodal input, function calling, and tool use, so the practical split is not just feature count. Use those differences to decide whether the page is about raw model quality, agentic coding support, multimodal ingestion, or predictable structured API behavior.

For cost, Mistral Large 2 lists $0.48/1M input and $2.4/1M output tokens, while Mistral Small 4 lists $0.15/1M input and $0.6/1M output tokens on the cheapest tracked provider. A 70/30 input-output blend puts Mistral Small 4 lower by about $0.77 per million blended tokens. Availability is 4 providers versus 3, so concentration risk also matters.

Choose Mistral Large 2 when vision-heavy evaluation and broader provider choice are central to the workload. Choose Mistral Small 4 when long-context analysis, larger context windows, and lower input-token cost are more important. For production, rerun your own prompts through the exact provider, region, and tool stack you plan to ship. This keeps the decision grounded in measurable tradeoffs instead of brand-level assumptions. It also helps separate model capability from provider packaging, which can change cost and latency.

FAQ

Which has a larger context window, Mistral Large 2 or Mistral Small 4?

Mistral Small 4 supports 256K tokens, while Mistral Large 2 supports 128K tokens. That gap matters most for long documents, large codebases, retrieval-heavy agents, and conversations where earlier context must remain visible.

Which is cheaper, Mistral Large 2 or Mistral Small 4?

Mistral Small 4 is cheaper on tracked token pricing. Mistral Large 2 costs $0.48/1M input and $2.4/1M output tokens. Mistral Small 4 costs $0.15/1M input and $0.6/1M output tokens. Provider discounts or batch pricing can still change the final bill.

Is Mistral Large 2 or Mistral Small 4 open source?

Mistral Large 2 is listed under True. Mistral Small 4 is listed under Apache 2.0. License labels affect whether you can self-host, redistribute weights, or rely only on hosted APIs, so confirm the upstream license before deployment.

Which is better for vision, Mistral Large 2 or Mistral Small 4?

Both Mistral Large 2 and Mistral Small 4 expose vision. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface.

Which is better for multimodal input, Mistral Large 2 or Mistral Small 4?

Both Mistral Large 2 and Mistral Small 4 expose multimodal input. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface.

Where can I run Mistral Large 2 and Mistral Small 4?

Mistral Large 2 is available on OpenRouter, IBM watsonx, AWS Bedrock, and Mistral AI Studio. Mistral Small 4 is available on OpenRouter, NVIDIA NIM, and Mistral AI Studio. Provider coverage can affect latency, region availability, compliance posture, and fallback options.

Continue comparing

Last reviewed: 2026-05-14. Data sourced from public model cards and provider documentation.