Skip to content
← Models
Open weightsMistral AI/10 trustVerified

Robostral Navigate

Mistral's 8B embodied navigation model: one RGB camera, plain-language instructions.

Visit official page →

About

Robostral Navigate is Mistral AI's first model for embodied navigation, introduced July 8, 2026. The 8B vision-language model steers robots through complex environments using only a single RGB camera plus a plain-language instruction, with no LiDAR, depth sensors or multi-camera rigs.

It scores 76.6% on the R2R-CE validation-unseen benchmark, beating the best single-camera approach by 9.7 points and the best multi-sensor system by 4.5 points despite using neither.

Trained entirely in simulation on roughly 400,000 trajectories across 6,000 virtual environments, it generalizes across wheeled, legged and flying robots. Weights are reported as Apache 2.0, making it a notable open physical-AI entry.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

July 8, 2026

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

roboticsnavigationembodied-aivision-language-action

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 116; Trending score 78/100 from researcher assessment.

Sources