xAIGrok 4.3
Best configuration ranks #81 of 610 on the BenchLeader Index at 60.7 ±6.0 (medium reasoning effort). Released 17 Apr 2026.
- Blended price
- $1.56/M
- $1.25 in · $2.50 out
- Output speed
- 113 tok/s
- First answer
- 12 s
- first token 0.51 s
- Context
- 1M
- Overall index61
- Agents & tools60
- Knowledge72
- Instruction following80
- Multimodal60
- Long context64
- Composite60
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 48.4 | #327 | 113 tok/s | 0.69 s | $0.0013 | Agents & tools 50 · Composite 47 · Instruction following 49 · Knowledge 48 · Long context 42 · Multimodal 49 |
| low | 59.5 | #102 | 114 tok/s | 5.06 s | $0.0013 | Agents & tools 57 · Composite 60 · Instruction following 78 · Knowledge 71 · Long context 63 · Multimodal 57 |
| mediumbest | 60.7 | #81 | 113 tok/s | 12 s | $0.0013 | Agents & tools 60 · Composite 60 · Instruction following 80 · Knowledge 72 · Long context 64 · Multimodal 60 |
| high | 50.1 | #293 | 118 tok/s | 21 s | $0.0013 | Agents & tools 32 · Coding 55 · Knowledge 51 · Maths 49 · Reasoning 64 |
| default | 57.8 | #136 | 118 tok/s | 21 s | $0.0013 | Agents & tools 67 · Coding 49 · Composite 37 · Human preference 53 · Instruction following 78 · Knowledge 65 · Long context 63 · Multimodal 60 · Reasoning 63 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | – | 88.8%#47 | – | Epoch AI Benchmarking Hub | |
| LMArena Hard Prompts | – | – | – | – | 1457#85 | LMArena | |
| LiveBench Reasoningnot in index | – | – | – | – | 70.8%#50 | LiveBench | |
| GPQA Diamond (AA)not in index | 65.8%#323 | 84.3%#142 | 89.0%#83 | – | 90.1%#64 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 6.8%#311 | 18.4%#174 | 30.0%#104 | – | 37.2%#70 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | 91.4%#22 | Vals AI |
Coding
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| SciCode | – | – | – | 47.3%#74 | – | SciCode | |
| WeirdML | – | – | – | – | 49.9%#66 | WeirdML | |
| LMArena Coding | – | – | – | – | 1488#81 | LMArena | |
| LMArena WebDev | – | – | – | – | 1357#89 | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | 69.9%#48 | LiveBench | |
| SciCode (AA)not in index | 39.4%#118 | – | – | – | 48.3%#78 | Artificial Analysis | |
| LiveCodeBench | – | – | – | 84.5%#37 | – | Vals AI | |
| IOI | – | – | – | 15.3%#29 | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | 71.4%#55 | – | Vals AI |
Agents & tools
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LiveBench Agentic Codingnot in index | – | – | – | – | 18.5%#52 | LiveBench | |
| Terminal-Bench Hard | 18.9%#161 | 26.5%#126 | 30.3%#110 | – | 37.9%#61 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 65.8%#161 | 88.9%#71 | 91.2%#59 | – | 97.7%#9 | Artificial Analysis | |
| Terminal-Bench 2.1 (Vals) | – | – | – | 42.0%#54 | – | Vals AI |
Maths
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | 42.8%#44 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | 14.6%#46 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | – | – | 93.3%#44 | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | 11.0%#49 | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | 84.3%#42 | LiveBench |
Knowledge
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | 33.2%#49 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | 55.8%#49 | LiveBench | |
| AA-Omniscience | -33.1#225 | 13.9#57 | 16.7#49 | – | 18.0#48 | Artificial Analysis | |
| MMLU-Pro | – | – | – | 85.8%#51 | – | Vals AI | |
| LegalBench | – | – | – | – | 84.5%#29 | Vals AI | |
| CorpFin | – | – | – | – | 68.5%#7 | Vals AI | |
| TaxEval | – | – | – | – | 70.8%#84 | Vals AI |
Instruction following
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 73.6%#44 | LiveBench | |
| IFBench | 47.6%#184 | 81.0%#6 | 83.3%#1 | – | 81.3%#5 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | – | 1443#76 | LMArena | |
| EQ-Bench 4 | – | – | – | – | 1075#24 | EQ-Bench |
Multimodal
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | – | 1228#63 | LMArena | |
| MMMU-Pro | 64.8%#148 | 72.8%#100 | 75.8%#67 | – | 78.1%#55 | Artificial Analysis |
Long context
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| AA-LCR | 32.3%#312 | 74.0%#120 | 75.0%#111 | – | 73.0%#132 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | 149.2#49 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | 62.3%#52 | LiveBench | |
| AA Intelligence Index | 14.5#229 | 24.3#124 | 24.8#119 | – | 25.4#114 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | 24.3#46 | – | Vals AI |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| SpaceXAI (ZDR) | 104 tok/s | 0.51 s | $1.25 | $2.50 | 1M | – |
| SpaceXAI | 90 tok/s | 0.57 s | $1.25 | $2.50 | 1M | – |
| SpaceXAI Priority | 65 tok/s | 0.51 s | $2.50 | $5.00 | 1M | – |
| SpaceXAI Priority (ZDR) | 21 tok/s | 0.48 s | $2.50 | $5.00 | 1M | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.200 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0013 | $0.0009 | 14.8 s |
| Summarise a 30-page report | 12,000 / 600 | $0.017 | $0.0071 | 17.5 s |
| Code edit | 6,000 / 1,500 | $0.011 | $0.0065 | 25.5 s |
| Agentic coding session | 60,000 / 4,000 | $0.085 | $0.038 | 47.7 s |
| Structured extraction | 2,000 / 200 | $0.0030 | $0.0014 | 13.9 s |
See also
Data as of 9 Sept 2026. Compare these configurations.