LLM API pricing / Mistral
What the Mistral API costs, model by model
Short answer. Mistral lists 8 of its own models with a per-token price on its API, and serves 3 made by Z.ai and DeepSeek. Input: from $0.10 (Ministral 3 3B) to $1.50 (Mistral Medium 3.5) per 1M tokens. Output: from $0.10 (Ministral 3 3B) to $7.50 (Mistral Medium 3.5). Read on Mistral's own pages between October 26 and October 30, 2026.
In plain words. You pay for the tokens you send (input) and the tokens the model writes (output). One request of 10,000 tokens in and 1,000 out costs $0.0011 on Ministral 3 3B and $0.0225 on Mistral Medium 3.5. Output costs more than input on 5 of the 8 models.
Paying for the app? See Mistral plans and prices →
Current models
8 models Mistral makes, cheapest input first, verified between October 26, 2026 and October 30, 2026. The last column is the cost of one request of 10,000 input and 1,000 output tokens at these prices, with no cache.
| Model | Input $/1M | Output $/1M | Cached input | Context | Example request | Verified |
|---|---|---|---|---|---|---|
| Ministral 3 3B | $0.10 | $0.10 | $0.01 | 256K | $0.0011 | 2026-10-26 |
| Ministral 3 8B | $0.15 | $0.15 | $0.015 | 256K | $0.0016 | 2026-10-26 |
| Mistral Small 4 | $0.15 | $0.60 | $0.015 | 256K | $0.0021 | 2026-10-26 |
| Ministral 3 14B | $0.20 | $0.20 | $0.02 | 256K | $0.0022 | 2026-10-26 |
| Codestral | $0.30 | $0.90 | $0.03 | 128K | $0.0039 | 2026-10-26 |
| Mistral Large 3 | $0.50 | $1.50 | $0.05 | 256K | $0.0065 | 2026-10-26 |
| Mistral Large 4 preview | $0.70 | $2.10 | $0.07 | 1M | $0.0091 | 2026-10-30 |
| Mistral Medium 3.5 | $1.50 | $7.50 | $0.15 | 256K | $0.0225 | 2026-10-26 |
Served by Mistral, made by Z.ai and DeepSeek
Mistral sells these models on its API at the prices below; another company makes them. They are not counted in the figures of Mistral's own models above.
| Model | Made by | Input $/1M | Output $/1M | Cached input | Context | Example request | Verified |
|---|---|---|---|---|---|---|---|
| GLM 5.3 Flash | Z.ai | $0.15 | $0.50 | $0.03 | 1M | $0.002 | 2026-10-26 |
| DeepSeek V4.1 Flash 1017 | DeepSeek | $0.30 | $1.20 | $0.007 | 1M | $0.0042 | 2026-10-30 |
| GLM 5.3 | Z.ai | $1.40 | $4.40 | $0.14 | 1M | $0.0184 | 2026-10-26 |
How Mistral describes its models
- Ministral 3 3B:
Ministral 3 3B is the smallest and most efficient model in the Ministral 3 family, offering robust language and vision capabilities in a compact package. Designed for edge deployment, it delivers high performance across diverse hardware, including local setups.
(docs.mistral.ai, read October 6, 2026) - Ministral 3 8B:
Ministral 3 8B is a powerful and efficient model in the Ministral 3 family, offering best-in-class text and vision capabilities. Built for edge deployment, it delivers high performance across diverse hardware, including local setups.
(docs.mistral.ai, read October 6, 2026) - Mistral Small 4:
Our powerful hybrid model unifying instruct, reasoning, and coding capabilities in a single model. 119B parameters with 6.5B active.
(docs.mistral.ai, read October 6, 2026) - Ministral 3 14B:
Ministral 3 14B is the largest model in the Ministral 3 family, offering state-of-the-art capabilities and performance comparable to its larger Mistral Small 3.2 24B counterpart. Optimized for local deployment, it delivers high performance across diverse hardware, including local setups.
(docs.mistral.ai, read October 6, 2026) - Codestral:
Our cutting-edge language model for code completion released end of July 2025, Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM) and code generation.
(docs.mistral.ai, read October 6, 2026) - Mistral Large 3:
Mistral Large 3, is a state-of-the-art, open-weight, general-purpose multimodal model with a granular Mixture-of-Experts architecture.
(docs.mistral.ai, read September 30, 2026) - Mistral Large 4:
Mistral Large 4 is a state-of-the-art, open-weight, general-purpose multimodal model with a granular Mixture-of-Experts architecture.
(docs.mistral.ai, read October 22, 2026) - Mistral Medium 3.5:
Our frontier-class multimodal model optimized for agentic and coding use cases.
(docs.mistral.ai, read September 30, 2026) - GLM 5.3 Flash (made by Z.ai):
Z.AI's GLM 5.3 Flash is a multimodal 320B-parameter (18B active) Mixture-of-Experts model. Built for developer agent workflows, visual coding, and complex tool-use tasks. The model is served without Mistral modifications.
(docs.mistral.ai, read October 23, 2026) - DeepSeek V4.1 Flash 1017 (made by DeepSeek):
DeepSeek V4.1 Flash 1017 is a third-party open weight text model from DeepSeek, hosted by Mistral. Ideal for long-context coding and agentic workflows. The model is served without Mistral modifications.
(docs.mistral.ai, read October 30, 2026) - GLM 5.3 (made by Z.ai):
A third-party open weight text model from Z.ai, hosted by Mistral for long-context coding and agentic workflows. The model is served without Mistral modifications.
(docs.mistral.ai, read October 6, 2026)
How Mistral prices its models
- 8 of the 8 models publish a cached-input price, at 10% of the input price.
What Mistral's own pages say
Preview
- Mistral Large 4:
October 6, 2026 Public Preview Mistral Ecosystem Open Weight License v 26.10
(docs.mistral.ai, read October 30, 2026) - Mistral's model lifecycle policy:
Are priced at the same rate as General Availability models.
(docs.mistral.ai, read October 30, 2026) - Mistral's model lifecycle policy:
Allow silent updates.
(docs.mistral.ai, read October 30, 2026) - Mistral's model lifecycle policy:
Have no guaranteed path to General Availability, and may be retired before reaching it.
(docs.mistral.ai, read October 30, 2026)
Aliases
- Mistral Large 3:
API names: `mistral-large-2512`, `mistral-large-latest`
(docs.mistral.ai, read October 26, 2026) - Mistral's model lifecycle policy:
model-name-latest | mistral-medium-latest | Latest version, across generations |
(docs.mistral.ai, read October 26, 2026) - Mistral's model lifecycle policy:
Using an alias may expose you to silent updates in model behavior and pricing.
(docs.mistral.ai, read October 26, 2026)
Deprecation notice
- Mistral's model lifecycle policy:
Public Preview | 1 month |
(docs.mistral.ai, read October 26, 2026) - Mistral's model lifecycle policy:
General Availability | 6 months |
(docs.mistral.ai, read October 26, 2026) - Mistral's model lifecycle policy:
Third-party models | 1 month |
(docs.mistral.ai, read October 26, 2026) - Mistral's model lifecycle policy:
During the deprecation period, the model remains accessible.
(docs.mistral.ai, read October 26, 2026) - GLM 5.2:
This model is deprecated. Use Z.ai GLM 5.3 for new integrations.
(docs.mistral.ai, read October 26, 2026) - GLM 5.2:
Deprecation date 9/29/2026
(docs.mistral.ai, read October 26, 2026)
Retirements
Deprecated, still callable: GLM 5.2 (made by Z.ai; no shutdown date announced).
Recorded price changes
- October 30, 2026 · price down 1.9×. Mistral cut the price of Mistral Large 4, in Public Preview, from $1.36 to $0.70 per million input tokens (1.36 ÷ 0.70 = 1.94×), from $4.18 to $2.10 for output (1.99×) and from $0.14 to $0.07 for cached input. The last reading at the old price was the weekly check of October 26, 2026, 'Mistral Large 4 ↗ | $1.36 | $0.14 | $4.18 |'; on October 30, 2026 the pricing page printed 'Mistral Large 4 ↗ | $0.7 | $0.07 | $2.1 |'. Mistral does not say when it changed the price or whether the new one is temporary, so the date here is the first day it was seen. source
- October 22, 2026 · correction of our own reading. Correction of our own reading, not a price change by Mistral. Mistral's pricing page prints a cached input price for each model in a 'Cached input' column, and seven rows of this dataset did not carry it: Mistral Large 3 ($0.05), Mistral Medium 3.5 ($0.15), Mistral Small 4 ($0.015), Ministral 3 14B ($0.02), Ministral 3 8B ($0.015), Ministral 3 3B ($0.01) and Codestral ($0.03) per million tokens, read on October 22, 2026. Before that the rows showed no cached input price; input and output prices are unchanged. When Mistral began printing these figures is not known, so the date here is the day they were read. source
- September 16, 2026 · added to this dataset, our omission: GLM 5.3. Full explanation
Analysis that cites Mistral models
- GLM 5.3 Flash API Pricing: $0.15 In, $0.50 Out, Same on Mistral and Z.ai · names 6 Mistral models
- AI Pricing This Week: OpenAI Shutdown, Mistral, Oct 19-25 · names 4 Mistral models
- Mistral Large 4 API Pricing: $1.36 In, $4.18 Out, in Public Preview · names 2 Mistral models
- Prompt Caching Is Not 10% Everywhere: What a Cache Hit Costs Across Eight Providers · names 2 Mistral models
- Mistral Large 4 Price Cut: $0.70 In, $2.10 Out, Still in Public Preview · names 1 Mistral model
- Mistral Pro (Le Chat) vs ChatGPT Plus: $14.99 or $20 · October 26, 2026
FAQ
What is the cheapest Mistral model on the API?
Ministral 3 3B, at $0.10 per 1M input tokens and $0.10 per 1M output tokens, as Mistral's pricing page printed it on October 26, 2026.
Which models does Mistral serve from other makers?
GLM 5.3 Flash (made by Z.ai, $0.15 in, $0.50 out per 1M tokens), DeepSeek V4.1 Flash 1017 (made by DeepSeek, $0.30 in, $1.20 out per 1M tokens) and GLM 5.3 (made by Z.ai, $1.40 in, $4.40 out per 1M tokens). Mistral's page for GLM 5.3 Flash says: “Z.AI's GLM 5.3 Flash is a multimodal 320B-parameter (18B active) Mixture-of-Experts model. Built for developer agent workflows, visual coding, and complex tool-use tasks. The model is served without Mistral modifications.”
Is Mistral Large 4 generally available?
Mistral Large 4: “October 6, 2026 Public Preview Mistral Ecosystem Open Weight License v 26.10”. Mistral's model lifecycle policy: “Are priced at the same rate as General Availability models.” “Allow silent updates.” “Have no guaranteed path to General Availability, and may be retired before reaching it.” Read on October 30, 2026.
What does the alias mistral-large-latest point to?
Mistral Large 3: “API names: `mistral-large-2512`, `mistral-large-latest`”. Mistral's model lifecycle policy: “model-name-latest | mistral-medium-latest | Latest version, across generations |”. “Using an alias may expose you to silent updates in model behavior and pricing.” Read on October 26, 2026.
How much notice does Mistral give before it retires a model?
Mistral's model lifecycle policy: “Public Preview | 1 month |”. “General Availability | 6 months |”. “Third-party models | 1 month |”. “During the deprecation period, the model remains accessible.” GLM 5.2: “This model is deprecated. Use Z.ai GLM 5.3 for new integrations.” GLM 5.2: “Deprecation date 9/29/2026”. Read on October 26, 2026.
How we checked
- Each price was read on Mistral's pricing page (docs.mistral.ai) or on a model page it links. The Verified column links to that page and gives the day.
- Every Monday a script reads Mistral's pricing page again and compares each row with the figures here. A difference is read by hand before a price changes.
- Quotes are copied as Mistral's pages print them, with the day they were read. A script checks every week that the model descriptions and the quotes under “What Mistral's own pages say” are still on their pages.
- The example request is worked out from the per-1M prices: 10,000 input tokens and 1,000 output tokens, no cache, no batch discount, no long-prompt tier.
Full method: methodology. All providers: the full pricing table.