Grok 4.1 Fast and Agent Tools API (xAI news) Nov 19, 2025
Grok 4.1 Fast
Grok 4.1 Fast ranks first on 2 of the 118 benchmarks with five or more models scored.
120
Benchmarks with a score
2
Ranked first, out of five or more models
4
In the top three
xAI Nov 2025
Grok 4.1 Fast
Closed weights
No Epoch Capabilities Index score yet
- τ³-bench Banking 13.1% #23 /25
- MathArena (Apex, IMO, USAMO) 5.21% #27 /47
- DeepPlanning 19.1% #12 /15
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| FRAMES | 87.6% | #1 of 9 | Lab model card | |
| MMLU-Pro | 84.2% | #35 of 212 | Official leaderboard | |
| FACTS Search | 77.4% | #7 of 31 | Official leaderboard | |
| LMArena (Elo) | 1,431 | #91 of 335 | Official leaderboard | |
| Extended NYT Connections | 87.4% | #30 of 103 | Official leaderboard | |
| KINA | 33.7% | #16 of 37 | Official leaderboard | |
| Judgemark v2.1 | 75.8 | #19 of 42 | Official leaderboard | |
| EQ-Bench 3 | 1,203 | #42 of 78 | Official leaderboard | |
| Creative Writing v3 | 61.4 | #80 of 120 | Official leaderboard | |
| BullshitBench V2 | 19% | #97 of 129 | Official leaderboard | |
| BullshitBench | 14.6% | #85 of 107 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 40.1 | #100 of 119 | Official leaderboard | |
| DesignArena ASCII Art | 1,102 | #66 of 77 | Official leaderboard | |
| SkateBench v2 | 35.4% | #18 of 21 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 17.8% | #95 of 104 | Official leaderboard | |
| Hemingway-bench | 916 | #40 of 42 | Official leaderboard | |
| Visual Aesthetic Benchmark | 15.5% | #18 of 18 | Official leaderboard | |
| HLE-Verified (text-only Full) | 29% | #6 of 6 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BRUMO 2025 | 97.5% | #8 of 42 | Official leaderboard | |
| HMMT Nov 2025 | 93.3% | #5 of 22 | Official leaderboard | |
| ForecastBench | 0.13 | #22 of 82 | Official leaderboard | |
| HMMT 2025 | 90% | #16 of 53 | Official leaderboard | |
| DTBench | 87.7% | #62 of 180 | Official leaderboard | |
| Echo General AI Prediction Leaderboard | 1,003 | #6 of 17 | Official leaderboard | |
| AIME 2025 | 89.2% | #23 of 53 | Official leaderboard | |
| SimpleBench | 56% | #51 of 106 | Official leaderboard | |
| CMIMC 2025 | 84.4% | #16 of 33 | Official leaderboard | |
| General365 | 53.1% | #12 of 23 | Official leaderboard | |
| SMT 2025 | 84.6% | #22 of 40 | Official leaderboard | |
| MathArena (Apex, IMO, USAMO) | 5.21% | #27 of 47 | Official leaderboard | |
| MathArena Apex Shortlist | 57.9% | #23 of 38 | Official leaderboard | |
| AIME 2026 | 94.2% | #20 of 31 | Official leaderboard | |
| IMO-ProofBench | 18.6% | #13 of 20 | Official leaderboard | |
| HMMT 2026 | 86.4% | #21 of 31 | Official leaderboard | |
| ProofBench | 4.00% | #53 of 59 | Official leaderboard | |
| SpatialBench (LatchBio) | 34% | #13 of 14 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| IOI (Vals) | 7.67% | #34 of 55 | Official leaderboard | |
| ALE-Bench | 1,111 | #92 of 126 | Official leaderboard | |
| DesignArena Website | 1,098 | #117 of 151 | Official leaderboard | |
| DesignArena Game Dev | 1,057 | #115 of 143 | Official leaderboard | |
| DesignArena Code | 1,082 | #116 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,044 | #111 of 136 | Official leaderboard | |
| Snorkel Agentic Coding | 25.2% | #9 of 11 | Official leaderboard | |
| DesignArena UI Component | 1,044 | #114 of 139 | Official leaderboard | |
| DesignArena Data Visualization | 1,086 | #107 of 118 | Official leaderboard | |
| DesignArena SVG | 999 | #94 of 103 | Official leaderboard | |
| Code Arena | 1,242 | #118 of 128 | Official leaderboard | |
| DesignArena Agentic Overall | 933 | #70 of 74 | Official leaderboard | |
| Vibe Code Bench | 1.20% | #115 of 120 | Official leaderboard | |
| WebDev Arena | 1,223 | #101 of 105 | WebDev Arena Leaderboard, via Epoch AI | |
| Vibe Code Bench v1 | 0% | #17 of 17 | Official leaderboard | |
| OTelBench | 2.90% | #14 of 14 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BFCL V4 | 69.6% | #5 of 65 | Official leaderboard | |
| tau2-bench Verified | 82.7% | #1 of 11 | Official leaderboard | |
| MCP-Universe | 40.7% | #3 of 30 | Official leaderboard | |
| Step Race | 4 | #8 of 59 | Official leaderboard | |
| LLM Chess | 800 | #32 of 123 | Official leaderboard | |
| Agent Poker Bench | 1,079 | #5 of 17 | Official leaderboard | |
| LisanBench | 1817.3 | #46 of 133 | Official leaderboard | |
| Cooperate to Compete (C2C) | 0.69 | #2 of 5 | Official leaderboard | |
| LLM-WikiRace | 5.00% | #10 of 21 | Official leaderboard | |
| Pencil Puzzle Bench | 3.30% | #26 of 53 | Official leaderboard | |
| BalatroBench | 7.67 | #8 of 16 | Official leaderboard | |
| LMArena Search Arena | 1,196 | #19 of 37 | Official leaderboard | |
| Clocktower Radio | 1,552 | #12 of 23 | Official leaderboard | |
| Game Arena Chess Openings | 635 | #6 of 11 | Official leaderboard | |
| AgencyBench | 44.3% | #5 of 9 | Official leaderboard | |
| PinchBench | 66.6% | #29 of 48 | Official leaderboard | |
| Elimination Game | 4 | #35 of 56 | Official leaderboard | |
| Game Arena Five in a Row | 373 | #14 of 20 | Official leaderboard | |
| Game Arena Chess | 607 | #31 of 42 | Official leaderboard | |
| GameWorld | 36% | #17 of 22 | Official leaderboard | |
| Game Arena Four in a Row | 198 | #34 of 43 | Official leaderboard | |
| DeepPlanning | 19.1% | #12 of 15 | Official leaderboard | |
| Game Arena Dark Hex | 126 | #33 of 41 | Official leaderboard | |
| ITBench | 39% | #11 of 13 | Official leaderboard | |
| Game Arena Checkers | 497 | #35 of 41 | Official leaderboard | |
| Game Arena Poker | -17.51 | #27 of 31 | Official leaderboard | |
| Game Arena Werewolf | -6 | #7 of 8 | Official leaderboard | |
| Game Arena Word Association | 37 | #36 of 40 | Official leaderboard | |
| τ³-bench Banking | 13.1% | #23 of 25 | Official leaderboard | |
| Kaggle Game Arena | 139 | #41 of 42 | Official leaderboard | |
| Research-Eval | 63.9% | #2 of 2 | Lab model card | |
| X Browse (xAI) | 56.3% | #2 of 2 | Lab model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CorpFin v2 | 66% | #25 of 119 | Official leaderboard | |
| Vals Finance Agent v1 | 44% | #13 of 47 | Official leaderboard | |
| TaxEval v2 | 73.1% | #50 of 129 | Official leaderboard | |
| CaseLaw v2 | 60.5% | #24 of 53 | Official leaderboard | |
| Vals Finance Agent | 52.5% | #23 of 49 | Official leaderboard | |
| LegalBench | 82.5% | #69 of 138 | Official leaderboard | |
| FailureSensorIQ | 64.6% | #12 of 24 | Official leaderboard | |
| MedScribe | 78.7% | #59 of 97 | Official leaderboard | |
| ProfBench | 44.8% | #46 of 72 | Official leaderboard | |
| Vending-Bench 2 | $1,107 | #53 of 67 | Official leaderboard | |
| EnterpriseBench: CoreCraft | 20.5% | #43 of 51 | Official leaderboard | |
| FoodTruck Bench | $817 | #50 of 59 | Official leaderboard | |
| Vending-Bench Arena | $723 | #28 of 32 | Official leaderboard | |
| PLawBench | 52.7% | #10 of 11 | Official leaderboard | |
| Vals Index v1 | 42.5% | #31 of 34 | Official leaderboard | |
| MortgageTax | 42.6% | #81 of 87 | Official leaderboard | |
| MedCode | 28.4% | #92 of 96 | Official leaderboard | |
| Public Benefits Bench | 44.8% | #48 of 50 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| WorldVQA | 21.1% | #8 of 22 | Official leaderboard | |
| Vision Arena | 1,196 | #82 of 136 | Official leaderboard | |
| OVO-S-Bench | 43.7% | #13 of 15 | Official leaderboard | |
| Math Kangaroo 2025 (MathArena) | 69% | #20 of 22 | Official leaderboard | |
| MineBench | 953 | #70 of 75 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PCT (Polarized Contrastive Pairs) | 32.5% | #4 of 46 | Official leaderboard | |
| Human Pathogen Capabilities Test | 49.7% | #48 of 86 | Official leaderboard | |
| Human Pathogen Capabilities Test v2 (HPCT v2) | 49.2% | #46 of 71 | Official leaderboard | |
| Molecular Biology Capabilities Test | 45.1% | #60 of 90 | Official leaderboard | |
| LiveSecBench | 32.7 | #24 of 34 | Official leaderboard | |
| Spiral-Bench v1.2 | 25.3 | #17 of 21 | Official leaderboard | |
| BinaryAudit | 7.02% | #19 of 20 | Official leaderboard | |
| KORA | 14.3% | #40 of 42 | Official leaderboard |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CAIS Risk Index | 60.0 | #52 of 67 | Official leaderboard | |
| CAIS Text Capabilities Index | 17.4 | #56 of 68 | Official leaderboard | |
| CAIS Vision Capabilities Index | 39.5 | #44 of 53 | Official leaderboard | |
| SAGE (Vals) | 31.3% | #73 of 81 | Official leaderboard | |
| SAGE v1.1 (Vals) | 74.5% | #27 of 29 | Official leaderboard |
Its lab's number next to a leaderboard
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| X Browse (xAI) | 56.3% X Browse (internal, Agent Tools API) | 56.3% Leaderboard, Agent Tools API | Same |
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.