GPT-6.1 Sol
OpenAI · released
The current mid-tier GPT: an upgraded Sol that OpenAI positions close to Astra on agentic coding, computer use and professional tasks at one-fifth of Astra's token prices, with unusually cheap cached input. It superseded GPT-6 Sol a week after that model launched.
Best for: Cost-efficient agentic coding · Computer use · Cache-heavy agents
- Intelligence Index
- 51.8
- #6 of 38 · at Max effort
- Price per 1M tokens
- $2 in $10 out
- Blended $4 at 3:1
- Context window
- 1.05M
- Up to 128K output
- Output speed
- 55 tokens/s
- $0.72 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.
- Overall84 · #6 of 38
Based on 3 of 3 benchmarks
- Coding85 · #6 of 39
Based on 5 of 5 benchmarks
- Agents & knowledge work70 · #13 of 40
Based on 6 of 6 benchmarks
- Reasoning & science92 · #5 of 39
Based on 2 of 2 benchmarks
- Knowledge & reliability92 · #3 of 39
Based on 2 of 2 benchmarks
- Vision93 · #2 of 31
Based on 2 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 |
|---|---|---|---|
| Max | 51.8 | $0.72 | 5.5 min |
| xHigh | 51.0 | $0.39 | 2.5 min |
| High | 50.2 | $0.32 | 61s |
| MediumSweet spot | 47.8 | $0.21 | 8.8s |
| Low | 42.1 | $0.13 | 3.4s |
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
#6 of 38· Max effort· Artificial Analysis
- Vals Index61.2%
#6 of 26· Max effort· $3.24 per test· Vals AI
- Arena · Text1483 ±11
#8 of 33· Max effort· 3,071 votes· Arena
Coding
- Terminal-Bench 4.056.1%
#5 of 38· Max effort· Artificial Analysis
- SciCode54.2%
#17 of 38· Max effort· Artificial Analysis
- Arena · WebDev1757 ±15
#4 of 34· Max effort· 1,965 votes· Arena
- Vibe Code Bench88.9%
#7 of 26· Max effort· Vals AI
- Code Migration63.3%
#5 of 26· Max effort· Vals AI
Agents & knowledge work
- GDPval-AA53.8
#16 of 39· Max effort· Artificial Analysis
- Arena · Agent+11.2 ±2.8
#5 of 29· Max effort· Arena
- Finance Agent v252.0%
#18 of 26· Max effort· Vals AI
#7 of 26· Max effort· Vals AI
- Legal Research Bench38.5%
#19 of 26· Max effort· Vals AI
- Tax Agent Bench62.3%
#17 of 26· Max effort· Vals AI
Reasoning & science
- Humanity's Last Exam52.9%
#6 of 39· Max effort· Artificial Analysis
- CritPt31.7%
#1 of 39· Max effort· Artificial Analysis
- ARC-AGI-2saturating94.2%
#2 of 16· Max effort· $0.25 per task· ARC Prize
- ARC-AGI-352.7%
#2 of 6· Max effort· ARC Prize
- GPQA DiamondsaturatingNot tested
Knowledge & reliability
#4 of 39· Max effort· Artificial Analysis
- AA-LCR83.0%
#9 of 39· Max effort· Artificial Analysis
Vision
- MMMU-Pro86.0%
#3 of 27· Max effort· Artificial Analysis
- Arena · Vision1291 ±16
#2 of 19· Max effort· 1,408 votes· Arena
Reported by OpenAI
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.
- DeepSWE v1.1 · High effort75.2%
- OSWorld 2.0 · offline set, partial score71.4%
Specifications
- Lab
- OpenAI
- Released
- Sep 29, 2026
- Availability
- Generally available
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
- Accepts
- Text, Images, PDFs
- Knowledge cutoff
- Apr 2026
- Weights
- Closed
- Reasoning control
- Effort low to max · pro mode
- API model ID
gpt-6.1-sol
Pricing
- Input / 1M tokens
- $2
- Cached input / 1M
- $0.10
- Output / 1M tokens
- $10
- Blended (3:1)
- $4
- Measured cost / task
- $0.72
Prompts over 272K input tokens are billed at $4 in / $15 out for the whole request. Batch and Flex are half price.
Official sources
Closest alternatives
The models nearest to this one on overall score.
23 benchmarks tracked. See methodology and sources.