Model
Model explorer

GPT-6 Sol

CLOSED
OpenAI · GPT-6 family · released Sep 22, 2026

API id gpt-6-sol, released 2026-09-22 together with GPT-6 Luna as cheaper GPT-6-generation tiers 'trained with similar methods as GPT-6 Astra'. It costs $2/$10 per Mtok, 50% below GPT-5.6 Sol's promotional $4/$20. It has a 1.05M context and 128K max output. It launched in ChatGPT Work/Codex and the API, and on 2026-10-07 became the default GPT-6 in ChatGPT for paid tiers. GPT-6.1 Sol superseded it in the API on 2026-09-29. On 2026-09-25 OpenAI fixed an image-encoding bug that had degraded GPT-6 Sol/Luna image understanding, so vision-dependent launch-week third-party scores may understate it. Most launch claims are cost-relative charts; only the four numeric figures below were printed.

ReasoningCodingVisionFunction callingTool useAgentic
2848.0
Elo · rank #12
Parameters
Undisclosed
Active params
—
Context
1.05M tokens
Architecture
Proprietary model; architecture and parameter count undisclosed; 1.05M-token context, selectable reasoning effort (none to max)
License
Proprietary
Languages
—
API price (in/out)
$2 / $10
Modalities
text · vision
Benchmark results
Bar shows position within the tracked field; marker = field best
AutomationBenchAgents33.2%#12
best: DeepSeek-V4.1-Flash · 54.8%
DeepSWECoding68.8%#11
best: Gemini 4 Argon · 77.9%
ExploitBenchAgents81.7%#3
best: GPT-6 Astra · 100.0%
FrontierCodeCoding49.3%#7
best: Claude Opus 5.5 · 54.6%
HealthBench ProfessionalKnowledge60.8%#5
best: Claude Opus 5.5 · 65.6%
OSWorld 2.0Agents60.5%#8
best: Claude Opus 5.5 · 81.8%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$2
Output / M tok
$10
API price $2/$10 · each benchmark row carries its own source badge (see methodology)