Mistral vs Together AI

Pricing, features, platforms and who each one suits — side by side.

Mistral

Open-weight AI models, chat and coding agent you can run on your own infrastructure

Free · from $14.99/moVisit site

Together AI

Cloud for running, fine-tuning and scaling open-source AI models on GPUs

PaidVisit site

Choose Mistral if…

you want a free plan to start with
you want open source you can self-host or audit
you work on iOS, Android, VS Code
you're in its core audience: developers, enterprises with data-control requirements

Choose Together AI if…

you're in its core audience: ai startups, ml engineers

Side by side

FeatureM

MistralT

Together AI

Pricing modelFreemiumPaid

Starting price$14.99/mo—

Free plan

Free trial

API

Open source

Trains AI on your contentYes, unless you opt out—

Data retentionVibe input/output kept until you delete conversation or account—

Optional cloud models—

OpenAI-compatible API—

Fine-tuning—

Speaker diarization—

Team admin / SSO—

PlatformsWeb, iOS, Android, API, VS Code, JetBrains, Self-hostedWeb, API

CategoryLocal & Open ModelsLLM APIs & Model Hosting

CompanyMistral AI · France · since 2023Together AI

Full feature lists

M Mistral

Open-weight models that can be self-hosted
Model customization on your own data with Forge
Terminal coding agent powered by Devstral
VS Code and JetBrains plugin with tab completion
Remote coding agents in isolated sandboxes
Custom MCP connectors and 100+ app connectors
Spreadsheet, slide and image generation in chat
Scheduled prompts, agents and workflows

T Together AI

Serverless pay-per-token inference for open-source models
Batch inference up to 30 billion tokens per model
Provisioned throughput with 99% uptime SLA
Dedicated single-tenant model and container endpoints
On-demand and reserved NVIDIA H100, H200, B200, B300 clusters
LoRA and full fine-tuning with SFT and DPO
Code sandboxes and code interpreter for agents
Managed storage with zero egress fees

Their home pages

What each vendor shows first: the pitch, the plans and the sign-up.

Mistral home page — M Mistral · mistral.ai

Together AI home page — T Together AI · together.ai

What you'd pay

The cheapest plan that fits, per month, billed yearly where the plan offers it.

Team size	Mistral	Together AI
1 person	$14.99 Pro	no published plan
5 people	$124.95 Team	no published plan
20 people	$499.80 Team	no published plan

What the free plans include

Mistral

Limited messages and web searches
Limited coding sessions
Basic image generation and editing
Limited libraries and document upload

Together AI

No free plan.

All plans and prices

Mistral

FreeFree

Pro$14.99/mo excluding taxes

Team$24.99/user/mo excluding taxes

EnterpriseCustom

Together AI

Serverless InferenceFrom $0.15 / $0.60 per 1M input/output tokens (gpt-oss-120B)

GPU Clusters (on-demand)$3.99/GPU-hour (H100)

Dedicated Inference$5.49/GPU-hour (H100)

Fine-tuningFrom $0.34 per 1M training tokens (Qwen3.5 9B SFT)

Provisioned ThroughputCustom

Price changes we've recorded

We re-read each pricing page every month and log what changed.

Mistral

No changes since we started tracking.

Together AI

Nov 8, 2026Serverless Inference: $0 → $0.15; New plan: GPU Clusters (on-demand) ($3.99); New plan: Dedicated Inference ($5.49); New plan: Fine-tuning ($0.34); Provisioned Throughput: Per PTU per minute → Custom; Plan dropped: Dedicated Inference (On-demand); Plan dropped: Dedicated Inference (Reserved)