whatmodel.sh / Grok 4.1 Fast (Reasoning)
Grok 4.1 Fast (Reasoning)
↗ share
+ contribute
Grok 4.1 Fast (Reasoning) is a closed-weights language model from SpaceXAI (xAI), released November 2025.
Catalogue identity notes
Comparison-only. xAI API model 'grok-4-1-fast-reasoning' (MathArena config xai/grok-41-fast-reasoning, date 2025-11-20). Distinct from the non-reasoning variant and from Grok 4.20/4.3/4.5.
$ how it compares
The LLMs ranked either side of Grok 4.1 Fast (Reasoning), out of 828 with a score, then more from SpaceXAI (xAI).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 141 |  DeepSeek V3.2 Exp | 67±1 | local | ||
| 142 |  Qwen3 235B-A22B Thinking 2507 | 66±1 | api $0.97/M +local | ||
| 143 | Grok 4.1 Fast (Reasoning) | 66±1 | api $0.28/M | ||
| 144 |  o4-mini | 66±1 | api $1.93/M | ||
| 145 | Qwen3.6 Flash | 66±1 | api $0.56/M | ||
| more from SpaceXAI (xAI) | |||||
| 30 | Grok 4.6 | 77±1 | api $3.00/M | ||
| 42 | Grok 4.5 | 75±1 | api $3.00/M | ||
| 47 | Grok 4.7 | 75±1 | api $3.00/M | ||
$ what it's good at
agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #110 of 300 · 17% of its score · 4 tests
Beats o3-pro, Qwen3.6 Plus, o3 here, though they rank higher overall. Trails SmolLM3 3B, Nova 2 Pro (Preview), Mistral Medium 3.5 (128B) here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| τ³-bench banking_knowledge (v1.0.1 grading) | 13.1% ✎ | ▃▃▂▃·▆▃▂█▅▅▂ beats 2/25 | 5.2% |
| BFCL V4 (December 2025 archive) | 69.6% ✎ | ▁▁▁▂██▅▅▆▂▃▂ beats 74/77 | 5.2% |
| Vibe Code Bench (test split, 50 web apps) | 1.2% ✎ | ▃··▃·▃██▃▃▃█ beats 0/14 | 4.0% |
| τ²-Bench-Verified (GPT-5.1 user) | 82.7% ✎ | ▂··▄···▂▄▄█▄ beats 13/13 | 2.1% |
coding▁▁▁▃▆██▇▆▄▃▁ #162 of 458 · 28% of its score · 3 tests
Beats Gemini 2.5 Pro, Qwen3-Coder-Next (80B-A3B) here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 8B Reasoning 2512, Phi-4-reasoning 14B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena WebDev (Code Arena) | 1241.3 ✎ | ▁▂▄▃▅█▄▇▆▄▁▁ beats 10/119 | 19% |
| WebApp1K-Duo-React pass@1 (onekq-ai) | 71.6% ✎ | ▁··▁·▁▁▁▆█▇█ beats 33/45 | 4.8% |
| Vibe Code Bench (test split, 50 web apps) | 1.2% ✎ | ▃··▃·▃██▃▃▃█ beats 0/14 | 4.0% |
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #78 of 393 * · 11% of its score · 1 test
Beats LongCat 2.0 (1.6T-A48B), Mistral Large 4 (public preview), Inkling Small (266B-A12B) here, though they rank higher overall. Trails Gemma 4 12B, Gemma 4 26B-A4B, Nemotron Cascade 2 30B-A3B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LMArena (text) | 1430.3 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 285/379 | 11% |
math▁▂▃▅▆▅▄█▄▁▁▁ #90 of 398 · 30% of its score · 4 tests
Beats Claude Sonnet 4.6, Claude Opus 4.5, MiniMax-M3 here, though they rank higher overall. Trails Gemini 2.5 Flash, Granite 4.2 30B, INTELLECT-3 here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| MathArena ArXivMath February 2026 (32 problems) | 32.0% ✎ | ▁▁▁▁█▇▄▃▁▃·▃ beats 8/24 | 12% |
| MathArena ArXivMath January 2026 (23 problems) | 53.3% ✎ | ▂··▄▂▄█▆▆▄▄█ beats 8/23 | 10% |
| MathArena ArXivMath December 2025 (17 problems) | 50.0% ✎ | ▃·▃▃█▁·▄·▁▁▁ beats 13/18 | 6.3% |
| FormalProofBench (200 graduate-level Lean 4 problems, agentic) | 4.0% ✎ | ▃▅·▃··▅▃█··▃ beats 1/10 | 1.6% |
vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #78 of 147 * · 7.3% of its score · 1 test
Beats DeepSeek V4 Flash Vision (Exp), GPT-5.4 nano, GPT-5 mini here, though they rank higher overall. Trails GPT-4.1 mini, Gemini 2.5 Flash, GPT-4.1 here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena vision (style controlled) | 1196.3 ✎ | ▁▁▂▂▂▂▄▅▅▆██ beats 68/145 | 7.3% |
web search▂▁▂▆▄█▃▃▁▁·▁ #32 of 80 * · 4.3% of its score · 1 test
Beats Gemini 3.1 Pro (Preview), GPT-5.3-Codex, Qwen3.7 Max here, though they rank higher overall. Trails GPT-4o mini (2024-07-18), Mistral Small 3.1 24B (2503), GPT-4o (2024-08-06) here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena search systems (style controlled) | 1194.2 ✎ | ▁·▁▁▂▂▂█▅▂▁▁ beats 17/30 | 4.3% |
other tests2 not in a task · 7.1% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| KINA (Knowledge Index of Noah's Ark) accuracy, 899 ten-option items | 33.7% ✎ | ▁▁▂▁▃▄█▆▄▅▂▃ beats 26/39 | 7.1% |
| Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate) | 80.8% ✎ | ▁▃▁▂▅█▂▃▄▃▁▁ beats 6/98 | 0.0% |
$ where to run it
Through an API
api · Microsoft Azure · $0.28/M · $0.2 in / $0.5 out per 1M tokens · 128k context · azure.microsoft.com ↗ · price sourceUSD per 1M tokens, Azure AI Foundry Models pay-as-you-go (xAI Grok table), 'Grok 4.1 Fast Global': input $0.20, output $0.50, region US East. The Azure model id is grok-4-1-fast-reasoning, version 1, GA, with no retirement date on Azure's schedule; the same row covers grok-4-1-fast-non-reasoning. Azure lists a 128,000-token input limit for this deployment. xAI's own price table no longer lists grok-4-1-fast-reasoning, and Google Cloud shut its Vertex endpoint down on 2026-08-20. Currency: USD. Microsoft Azure official pricing page (price cells carry the amounts in the page's own data), observed 2026-09-26.
$ how the score was worked out · LMArena leaderboard dataset 41%, MathArena ArXivMath / BrokenA… 28%, Knowledge Index of Noah's Ark… 7% and 7 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 LMArena leaderboard dataset | 4 | 41% |
 MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets | 3 | 28% |
 Knowledge Index of Noah's Ark (arXiv:2606.05104v2) | 1 | 7% |
 τ²/τ³-bench leaderboard: Sierra's own evaluation runs | 1 | 5% |
| Berkeley Function Calling Leaderboard result archive | 1 | 5% |
| WebApp1K models leaderboard (onekq-ai) | 1 | 5% |
| Vibe Code Bench (Vals AI, arXiv:2603.04601v3) | 1 | 4% |
| Amazon AGI — τ²-Bench-Verified | 1 | 2% |
| FormalProofBench (Vals AI, arXiv:2603.26996v1) | 1 | 2% |
| Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository) | 1 | 0% |
| prior · pull toward the middle | 1% | |
| weighted average | = 66.4 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json