Model comparison
Compare Mistral Medium 3.5 and Kimi K3 using the same provider-sourced text & reasoning rubric. No mystery score and no invented benchmark ranking.
Facts checked September 4, 2026
Set your usage. Your estimate updates as you type.
Example: 6,000 input tokens for 10 pages, plus a 500-token summary. Page lengths vary; adjust the numbers below.
One run sends your input to the model once and receives an answer. Tokens are pieces of text: input is what you send, output is the answer you receive.
Estimates exclude taxes, tools, cache storage/writes, free allowances and custom discounts. Image estimates cover output only, not prompt or reference-image charges. Quality modes differ by model. Unlisted settings are not treated as free.
| Model | Access | Estimated total (USD) |
|---|---|---|
| Kimi K3Moonshot AI | Moonshot AI | No reviewed rate |
| Mistral Medium 3.5Mistral AI | Mistral AI | No reviewed rate |
Quick take
Mistral's open-weight frontier model for demanding multimodal, coding, reasoning, and agent workflows, with a 256K context window.
The headline token price applies to Mistral's standard hosted endpoint. Regional inference costs more, and self-hosting shifts the bill to GPUs, operations, and support.
Moonshot AI's frontier multimodal model for million-token coding, reasoning, knowledge work, and tool-driven agents.
Moonshot's public terms permit broad service-improvement uses of submitted content. Treat the provider route and enterprise agreement as a core requirement for confidential work.
Compare the published facts
Values use each provider's own published units and limits. A blank means the provider did not publish a directly comparable value in the sources reviewed.
| Text & reasoning | Mistral Medium 3.5 | Kimi K3 |
|---|---|---|
| Context windowMaximum combined prompt and working context documented by the provider. | 256K tokens combined | 1M tokens |
| Maximum outputProvider-published response limit, where available. | Not separately published | Not separately published for the direct API |
| Knowledge cutoffLatest reliable knowledge date explicitly published by the model provider. Search and connected tools can retrieve newer information but do not change the model's built-in cutoff. | Not published | Not published |
| Input priceCurrent standard list price per million input tokens unless noted. | $1.50 / 1M | $3 / 1M uncached; $0.30 cached |
| Output priceCurrent standard list price per million output tokens unless noted. | $7.50 / 1M | $15 / 1M |
| InputsMedia types accepted by the listed model endpoint. | Text, image | Text, image |
| Tools & agentsSelected native tools and agent-building capabilities, not an exhaustive list. | Functions, agents, built-in tools, structured and predicted outputs | Tools, agent workflows, reasoning levels, dynamic tool loading on supported routes |
How to choose
Start with the job you need to complete, then validate cost, access, and policy details on your exact provider route.
Mistral's commercial terms exclude model training by default except when a customer opts in or uses designated Labs or preview models. Standard API input and output retention is generally 30 rolling days for abuse monitoring unless zero data retention is enabled.
Provider and API links
Moonshot's API privacy and model-use terms allow submitted content to be stored and used to provide, maintain, develop, and improve the service. Obtain appropriate enterprise terms before sending sensitive material.
Frequently asked questions
Mistral Medium 3.5: Mistral's open-weight frontier model for demanding multimodal, coding, reasoning, and agent workflows, with a 256K context window. Kimi K3: Moonshot AI's frontier multimodal model for million-token coding, reasoning, knowledge work, and tool-driven agents.
Consider Mistral Medium 3.5 when your priority is Complex coding and agent workflows. Consider Kimi K3 when your priority is Large codebase and document work. Test both with your own data and provider route before committing.
No. This comparison aligns provider-published facts for the Text & reasoning category. It does not claim a universal winner or combine incompatible third-party benchmark scores.
