Tesla FSD 14.3.9 Adds Automatic Collision Evasion

According to the description, the new evasion assistance uses Full Self-Driving (Supervised) to guide the vehicle safely and keep it stable in a potential crash situation, then continue driving afterward. The key point is that it can intervene even when you are driving manually.

Tesla Lists These Two Scenarios

• A head-on collision is imminent and braking alone is not expected to be sufficient to avoid the impact.
• The vehicle detects insufficient attention (for example, the driver’s gaze and posture are not directed toward the road) or a possible unintentional deactivation of FSD (Supervised).

In everyday driving, this means that if a sudden obstacle appears or a situation escalates, Tesla can use an AI-assisted evasion strategy in addition to ABS and electronic stability control. This could provide a genuine safety benefit, but much depends on how reliably the system recognizes objects, lanes and open space.

One point remains crucial: Even with FSD (Supervised), you are responsible and must remain attentive at all times. The system does not make your car autonomous.

FSD (Supervised) 14.3.9: What Do the Release Notes Say Has Improved?

Tesla lists a broad set of changes for v14.3.9, ranging from training methods to the runtime system. The clear focus is on greater robustness in difficult or rare scenarios and faster responses.

AI Training and Perception

Tesla cites an upgrade to the reinforcement learning phase (RL), which is intended to improve performance in many driving situations. The neural network’s vision encoder has also been revised, including improvements for rare events and poor visibility, a stronger understanding of 3D geometry and expanded traffic-sign capabilities.

20% Faster Response Time With a New AI Compiler

One particularly technical detail: Tesla says it has fundamentally rebuilt the AI compiler and runtime using MLIR. According to the release notes, the result is an approximately 20% faster response time and faster model iteration. In practical terms, this means lower latency between perception and action—precisely where driving behavior can feel more “human” and safer.

Specific Behavior and Comfort Improvements

Which Cars Will Get It, and What Hardware Is Required?

The release notes assign the new features to HW4. The affected models listed are the Model S, Model 3, Model X, Model Y and Cybertruck, with the collision-evasion feature also referencing HW4.

For the DACH region, the crucial point is that even if a vehicle has the appropriate hardware, the scope and timing of FSD features depend heavily on local regulatory approval. Experience shows that many of the capabilities described are not available in Europe on a one-to-one basis with the United States.

Assessment: What Does This Mean for Drivers in Everyday Use?

Automatic Collision Evasion may sound like a small feature, but it could have a major effect in practice: braking is not always enough, and evasive steering is often the last chance. At the same time, this is an especially challenging task because the system must decide within milliseconds whether the space to the right or left is truly clear and what secondary consequences could arise.

Tesla’s use of FSD (Supervised) as the “engine” for this feature fits with the release notes’ claim that the AI runtime’s response time has improved by 20%. Faster perception and action can make the difference in critical moments, provided that the system continues to detect its surroundings reliably.