When a startup scales its chatbot, the difference between a $0.019 per 1M input token from Mistral’s nemo and a $0.065 per 1M input token from DeepSeek’s flash‑0731 can add up quickly. Mistral’s lower‑tier models stay under $0.2 per 1M output, while DeepSeek’s flash tier pushes to $0.18 per 1M output, and higher‑end Chat models climb to $0.95 or $1 per 1M output.
Token‑by‑token price sheet
| Vendor | Model | Input | Cached input | Output |
|---|---|---|---|---|
| Mistral | mistral‑nemo | $0.019 per 1M tokens | – | $0.03 per 1M tokens |
| Mistral | mistral‑small‑24b‑instruct‑2501 | $0.05 per 1M tokens | – | $0.08 per 1M tokens |
| Mistral | mistral‑small‑3.2‑24b‑instruct | $0.075 per 1M tokens | – | $0.2 per 1M tokens |
| Mistral | ministral‑3b‑2512 | $0.1 per 1M tokens | $0.01 per 1M tokens | $0.1 per 1M tokens |
| Mistral | voxtral‑small‑24b‑2507 | $0.1 per 1M tokens | $0.01 per 1M tokens | $0.3 per 1M tokens |
| Mistral | mistral‑small‑2603 | $0.15 per 1M tokens | $0.015 per 1M tokens | $0.6 per 1M tokens |
| Mistral | ministral‑8b‑2512 | $0.15 per 1M tokens | $0.015 per 1M tokens | $0.15 per 1M tokens |
| Mistral | ministral‑14b‑2512 | $0.2 per 1M tokens | $0.02 per 1M tokens | $0.2 per 1M tokens |
| Mistral | mistral‑saba | $0.2 per 1M tokens | $0.02 per 1M tokens | $0.6 per 1M tokens |
| Mistral | codestral‑2508 | $0.3 per 1M tokens | $0.03 per 1M tokens | $0.9 per 1M tokens |
| DeepSeek | deepseek‑v4‑flash‑0731 | $0.065 per 1M tokens | $0.016 per 1M tokens | $0.18 per 1M tokens |
| DeepSeek | deepseek‑v4‑flash | $0.0668 per 1M tokens | $0.0134 per 1M tokens | $0.1336 per 1M tokens |
| DeepSeek | deepseek‑v4.1‑flash | $0.15 per 1M tokens | $0.003 per 1M tokens | $0.6 per 1M tokens |
| DeepSeek | deepseek‑v4‑flash‑vision‑exp | $0.22 per 1M tokens | $0.007 per 1M tokens | $0.66 per 1M tokens |
| DeepSeek | deepseek‑chat‑v3.1 | $0.25 per 1M tokens | $0.13 per 1M tokens | $0.95 per 1M tokens |
| DeepSeek | deepseek‑chat‑v3‑0324 | $0.25 per 1M tokens | – | $1 per 1M tokens |
| DeepSeek | deepseek‑chat | $0.2574 per 1M tokens | – | $1.0287 per 1M tokens |
| DeepSeek | deepseek‑v3.2 | $0.269 per 1M tokens | $0.1345 per 1M tokens | $0.4 per 1M tokens |
| DeepSeek | deepseek‑v3.2‑exp | $0.27 per 1M tokens | – | $0.41 per 1M tokens |
| DeepSeek | deepseek‑v3.1‑terminus | $0.27 per 1M tokens | $0.135 per 1M tokens | $1 per 1M tokens |
Mistral’s lower‑tier models are ideal for teams that need to keep per‑token costs near $0.02–$0.08 for input and $0.03–$0.08 for output, while DeepSeek offers a slightly higher starting point but scales to $0.66–$1.0287 per 1M output for its chat‑optimized tiers. If your budget prioritizes the absolute cheapest input and you can tolerate a modest output fee, Mistral’s nemo or small‑24b‑instruct‑2501 are the way to go. For applications that demand higher output quality and can afford the premium, DeepSeek’s chat models provide a richer experience at $0.95–$1.0287 per 1M output.
Pricing verified on 2026-09-12.