AI Model Watch
Models / Mistral

Leanstral 1.5

Preview

Specialist model for Lean 4 formal proof engineering, automated theorem proving and autoformalization -> not a general chat model. Labs tier, sparse MoE 119B total / 6.5B active. FREE during the Labs preview: pricing.free=true, $0/M in and out per the official docs model schema. Open weights: Apache 2.0 (huggingface.co/mistralai/Leanstral-1.5-119B-A6B). Text + image in, text out; 256K context, 128K max output. Mistral marks it status Active in its experimental Labs tier with a stated retirement of 2026-09-30 -> we normalize Labs/experimental to preview (it is not a GA production model), which also keeps the catalog's first $0 model out of the GA 'cheapest' rankings where a free specialist would rank misleadingly. Knowledge cutoff not published -> null.

Leanstral 1.5 by Mistral costs Free per 1M input tokens and Free per 1M output tokens (Free/1M blended), with a 256K (256.000-token) context window. It is scheduled for retirement, shutting down 30 Sep 2026. Its weights are publicly downloadable, so it can be self-hosted on your own infrastructure.

Last verified: 26 Sep 2026 · sourced from official provider documentation

Retires on 30 Sep 2026

Lifecycle warning

This model is deprecated.

Provider
Mistral
Status
Preview
Input price
Free / 1M tokens
Output price
Free / 1M tokens
Cached input
—
Blended price
Free / 1M tokens
Context window
256.000 tokens (256K)
Max output
128.000 tokens
Modality
text, vision
Open weights
Yes — self-hostable
Knowledge cutoff
—
Released
30 Jun 2026
API string
labs-leanstral-1-5

Source: Mistral official documentation ↗

See also

Explore more

All Mistral models→

Prices, context windows and lifecycle status for every Mistral model we track.

Compare models head-to-head→

Side-by-side price, context and status for the models developers weigh against each other.

FAQ

Leanstral 1.5 — questions & answers

How much does Leanstral 1.5 cost?

Leanstral 1.5 is priced at Free per 1M input tokens and Free per 1M output tokens (Free/1M blended at a 3:1 input-to-output ratio).

What is the context window of Leanstral 1.5?

Leanstral 1.5 has a 256.000-token context window (256K), with up to 128.000 output tokens per request.

Is Leanstral 1.5 being deprecated or retired?

Leanstral 1.5 is scheduled for retirement. Its scheduled shutdown date is 30 Sep 2026.

Does Leanstral 1.5 support image or vision input?

Yes — Leanstral 1.5 accepts image input. Listed modalities: text, vision.

Can I self-host Leanstral 1.5?

Yes — Leanstral 1.5 is an open-weight model: its weights are publicly downloadable, so you can run it on your own infrastructure (subject to its license) instead of only calling Mistral's API.

What is the API model string for Leanstral 1.5?

The API model identifier for Leanstral 1.5 is "labs-leanstral-1-5" when calling Mistral's API.