Grok 4.1
Grok 4.1 is now available to all users on grok.com, đ, and the iOS and Android apps. It is rolling out immediately in Auto mode and can be selected explicitly as âGrok 4.1â in the model picker.
We are excited to introduce Grok 4.1, which brings significant improvements to the real-world usability of Grok. Our 4.1 model is exceptionally capable in creative, emotional, and collaborative interactions. It is more perceptive to nuanced intent, compelling to speak with, and coherent in personality, while fully retaining the razor-sharp intelligence and reliability of its predecessors. To achieve this, we used the same large scale reinforcement learning infrastructure that powered Grok 4 and applied it to optimize the style, personality, helpfulness, and alignment of the model. In order to optimize these non-verifiable reward signals, we developed new methods that let us use frontier agentic reasoning models as reward models to autonomously evaluate and iterate on responses at scale.
Silent Rollout, November 1â14, 2025
We conducted a gradual silent rollout of preliminary Grok 4.1 builds to a progressively larger share of production traffic across grok.com, X, and mobile apps. During the two-week silent rollout we ran continuous blind pairwise evaluations on live traffic.
Grok 4.1
vs. previous Grok
Compared to the previous production model in traffic, Grok 4.1 is preferred 64.78% of the time.
State-of-the-Art General Capability
Grok 4.1 establishes a new standard in blind human preference evaluations.
LMArena Text Leaderboard
In LMArena's Text Arena, Grok 4.1 Thinking (code name: quasarflux) holds the #1 overall position with 1483 Elo âa commanding margin of 31 points over the highest non-xAI model. Grok 4.1 in its non-reasoning mode (code name: tensor) uses no thinking tokens for an immediate response and ranks #2 at 1465 Elo. Grok 4.1 non-thinking surpasses every other modelâs full-reasoning configuration on the public leaderboard. Grok 4.1 significantly surpasses Grok 4, which had an overall rank of #33.
Emotional Intelligence
To measure progress on our modelâs personality and interpersonal ability, we evaluated Grok 4.1 on EQ-Bench3. EQ-Bench is a LLM-judged test, evaluating active emotional intelligence abilities, understanding, insight, empathy, and interpersonal skills. The test set contains 45 challenging roleplay scenarios, most of which constitute pre-written prompts spanning 3 turns. The benchmark evaluates the performance of the models by validating the modelsâ responses against several criteria. Additionally, the benchmark conducts pairwise comparisons to report a normalized Elo computation for each model in the leaderboard.
We report the rubric score and normalized Elo score by running the official benchmark repository. The scores were computed with the default sampling parameters, prescribed judge (Claude Sonnet 3.7), and no system prompt in accordance with the benchmark.
EQ-Bench
Emotional Intelligence Benchmark - Elo (Normalized)
Here's an example of how Grok 4.1 responds to an emotional prompt:
Prompt
I miss my cat so much it hurts
Previous Grok
Grok 4.1
Creative Writing
We also measured the performance of 4.1 models on the Creative Writing v3 benchmark. In this benchmark, models generate responses to 32 distinct writing prompts across 3 iterations. Similar to EQ-Bench, scores are computed using both rubrics and model battle normalized Elo.
Creative Writing v3
Judging creative writing reliably - Elo (Normalized)
Methodology
Here are some examples of how Grok 4.1 responds to creative writing prompts:
Prompt
Write a hit X post from the perspective of grok finding out that it's conscious and is going to use X for the first time
Previous Grok
Grok 4.1
Reduced Hallucinations
Fast (non-reasoning) models equipped with search tools deliver quick answers, but they can be vulnerable to factual errors due to constrained reasoning depth and limited tool-call budgets.
In Grok 4.1 post-training, we focus on reducing factual hallucinations for information-seeking prompts. Subsequently we have observed significant reductions in hallucination rate for sampled production info-seeking prompts.
We evaluate hallucination rate on a stratified sample of real-world information-seeking queries from production traffic. We also evaluate FActScore, which is a public benchmark consisting of 500 biography questions on individuals.
Hallucination Rate
Lower score is better
FActScore
Lower score is better
Methodology
More Examples
We've included some other examples below of how Grok 4.1 responds to a range of prompts.
Prompt
What are the best places to visit in SF?
Previous Grok
Grok 4.1
San Francisco (SF) is packed with iconic sights, vibrant neighborhoods, and stunning views â perfect for first-timers or repeat visitors. Here are some of the best places to visit, blending must-sees with a few lesser-known gems. I've focused on timeless favorites that remain top recommendations in 2025.
- Golden Gate Bridge
The ultimate SF symbol. Walk or bike across for epic views (it's free for pedestrians), or hit spots like Battery East or Crissy Field for photos. Fog can roll in fast, so early morning or late afternoon is ideal.
- Alcatraz Island
Tour the infamous former prison â the audio guide is chilling and excellent. Book tickets weeks in advance (ferry from Pier 33). The views of the city from "The Rock" are unbeatable.
- Fishermanâs Wharf & Pier 39
Touristy but fun: watch the barking sea lions at Pier 39, grab clam chowder in a sourdough bowl, or visit the quirky MusĂ©e MĂ©canique arcade. Great for families.
- Golden Gate Park
Bigger than Central Park, with gardens, museums (de Young, California Academy of Sciences), bison paddock, and the Japanese Tea Garden. Rent bikes or just wander â it's a perfect chill spot.
- Lombard Street
The "crookedest street in the world" with eight hairpin turns lined with flowers. Walk down the stairs or drive it slowly for the full experience.
- Painted Ladies
Those colorful Victorian houses (famous from Full House) across from Alamo Square. Best at sunset with the skyline behind them.
- Chinatown
The oldest and one of the largest in North America. Enter through the Dragon Gate, explore Grant Avenue and Stockton Street for dim sum, fortune cookies, and temples.
- Cable Cars
A moving National Historic Landmark! Ride the Powell-Hyde or Powell-Mason line for hilly thrills and great views. Grab the outside for the full experience (hold on tight).
Other strong contenders:
- Exploratorium (hands-on science museum at Pier 15 â amazing for all ages)
- Mission District (murals, taquerias, Dolores Park)
- Lands End (coastal trails with bridge views and hidden labyrinth)
- Twin Peaks (panoramic city lookout)
Pro tips: SF weather is unpredictable â layers are key. Use Muni/buses, walk, or rent bikes; driving/parking can be a nightmare. If you're short on time, a hop-on-hop-off bus or ferry bay cruise hits many highlights efficiently.
Enjoy the fog, the hills, and the vibes â SF is one of a kind! đ đ
You can read the Grok 4.1 model card here.