Grok 4.7
SpaceXAI · released
The current top Grok model. It always reasons and is tuned for long-running coding agents and professional document work, at a list price well below the other flagships it is compared with.
Best for: Agentic coding · Long multi-step tasks · Tool-using agents
- Intelligence Index
- 46.4
- #8 of 38 · at xHigh effort
- Price per 1M tokens
- $2 in $6 out
- Blended $3 at 3:1
- Context window
- 500K
- Output speed
- 73 tokens/s
- $3.74 measured cost per task
Capability profile
One 0–100 score per area, placing this model between the weakest and strongest tracked model across that area's benchmarks.
- Overall69 · #13 of 38
Based on 3 of 3 benchmarks
- Coding61 · #12 of 39
Based on 5 of 5 benchmarks
- Agents & knowledge work73 · #8 of 40
Based on 6 of 6 benchmarks
- Reasoning & science60 · #16 of 39
Based on 2 of 2 benchmarks
- Knowledge & reliability81 · #11 of 39
Based on 2 of 2 benchmarks
- Vision69 · #14 of 31
Based on 1 of 2 benchmarks
Reasoning effort trade-off
The same model at each effort setting. Lower settings give up a few points for a fraction of the cost and wait.
| Setting | Intelligence Index | Cost / task | First answer |
|---|---|---|---|
| xHigh | 46.4 | $3.74 | 74s |
| High | 46.3 | $2.73 | 87s |
| LowSweet spot | 42.2 | $1.25 | 5.1s |
Measured by Artificial Analysis. First answer is the median wait until the first answer token, including thinking time. “Sweet spot” is the cheapest setting that keeps at least 90% of the top score for no more than 60% of its cost.
Benchmark results
All third-party results on record. Rank is among the tracked models that have been tested on that benchmark.
Overall
#8 of 38· xHigh effort· Artificial Analysis
- Vals Index55.0%
#9 of 26· xHigh effort· $12.1 per test· Vals AI
- Arena · Text1442 ±8
#24 of 33· xHigh effort· 6,405 votes· Arena
Coding
- Terminal-Bench 4.025.8%
#17 of 38· xHigh effort· Artificial Analysis
- SciCode57.4%
#11 of 38· xHigh effort· Artificial Analysis
- Arena · WebDev1638 ±11
#10 of 34· xHigh effort· 3,734 votes· Arena
- Vibe Code Bench86.2%
#8 of 26· xHigh effort· Vals AI
- Code Migration43.2%
#14 of 26· xHigh effort· Vals AI
Agents & knowledge work
- GDPval-AA60.8
#4 of 39· xHigh effort· Artificial Analysis
- Arena · Agent+4.0 ±1.6
#8 of 29· XHigh effort· Arena
- Finance Agent v252.3%
#17 of 26· xHigh effort· Vals AI
#10 of 26· xHigh effort· Vals AI
- Legal Research Bench47.1%
#8 of 26· xHigh effort· Vals AI
- Tax Agent Bench65.6%
#10 of 26· xHigh effort· Vals AI
Reasoning & science
- Humanity's Last Exam43.1%
#14 of 39· xHigh effort· Artificial Analysis
- CritPt17.7%
#17 of 39· xHigh effort· Artificial Analysis
- ARC-AGI-2saturating61.4%
#10 of 16· xHigh effort· $2.01 per task· ARC Prize
- ARC-AGI-31.8%
#4 of 6· xHigh effort· ARC Prize
- GPQA DiamondsaturatingNot tested
Knowledge & reliability
#6 of 39· xHigh effort· Artificial Analysis
- AA-LCR76.7%
#30 of 39· xHigh effort· Artificial Analysis
Vision
- MMMU-Pro78.2%
#14 of 27· Low effort· Artificial Analysis
- Arena · VisionNot tested
Reported by SpaceXAI
Numbers the lab published at launch, run on its own harness and settings. Useful context, but not independently verified and never used in the rankings on this site.
- Terminal-Bench 4.0 · xHigh effort37.6%
Specifications
- Lab
- SpaceXAI
- Released
- Sep 21, 2026
- Availability
- Generally available
- Context window
- 500,000 tokens
- Max output
- —
- Accepts
- Text, Images
- Knowledge cutoff
- May 2026
- Weights
- Closed
- Reasoning control
- Always on · effort low to xhigh
- API model ID
grok-4.7
Pricing
- Input / 1M tokens
- $2
- Cached input / 1M
- $0.50
- Output / 1M tokens
- $6
- Blended (3:1)
- $3
- Measured cost / task
- $3.74
Prompts of 200K tokens or more are billed at $4 in / $12 out for the whole request.
Worth knowing
- The documentation states no fixed output limit.
Official sources
Closest alternatives
The models nearest to this one on overall score.
23 benchmarks tracked. See methodology and sources.