Devstral 2

Devstral 2 is Mistral's coding-agent model for repository work, terminal tasks, and software fixes. It is the latest in the devstral line, following Devstral Small. It has a context window of 262144 tokens and a maximum output of 262144 tokens. It accepts and returns text, supports tool calling, and does not have a reasoning mode. The weights are open. Pricing starts at $0.40 per million input tokens and $2.00 per million output tokens, with cached input at $0.04 per million tokens.

Against Devstral Small: window 128K to 262K, input $0.10 to $0.40, output $0.30 to $2.00.

Range across the hosts that serve it: up to $0.478 per 1M input tokens.

• ·262k token context window
• ·262k token max output
• ·Tool calling for agent workflows
• ·Open weights for self-hosting
• ·Served by 10 hosts

• Reach for it for repository-level coding tasks
• Reach for it for terminal command execution
• Reach for it for software bug fixes
• Reach for it for agentic workflows with tool use

10 hosts serve this model at the price above · the maker's documentation

The registry keeps the six most recent versions of a line. Older ones stay in the file and can be surfaced without a rebuild.

What is the context window?

The context window is 262144 tokens, and the maximum output is also 262144 tokens.

Does it support tool calling?

Yes, it supports tool calling. It does not have a reasoning mode.

What is the price?

Input starts at $0.40 per million tokens, output at $2.00 per million tokens. Cached input is $0.04 per million tokens.