Model
Model explorer

GPT-6.1 Sol

CLOSED
OpenAI · GPT-6.1 family · released Sep 29, 2026

API id gpt-6.1-sol, released 2026-09-29 (DevDay week) as 'an upgrade to GPT-6 Sol that nearly matches GPT-6 Astra’s intelligence on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices'. It costs $2/$10 per Mtok with cached input at $0.10 (half of GPT-6 Sol's). It has a 1.05M context and 128K max output, and it is now the 'Default' Sol on the API models page. The launch post gives benchmark results only as charts and relative deltas: DeepSWE v1.1 is +6.4 pp over GPT-6 Sol's best, AutomationBench 1.0.6 is +2.2 pp over Opus 5.5 at medium and +4.8 pp over GPT-6 Sol, OSWorld 2.0 offline is +7 pp over GPT-6 Sol at max and within 2.1 pp of Astra, and Terminal-Bench Science 0.1 is more than double GPT-6 Sol at max, at $5.47/task vs $23.21 for Opus 5.5. None of these can be filed as absolute scores. OpenAI applies Astra-level cyber safeguards and a phased Daybreak rollout.

ReasoningCodingVisionFunction callingTool useAgentic
3125.4
Elo · unrated
Parameters
Undisclosed
Active params
—
Context
1.05M tokens
Architecture
Proprietary model; architecture and parameter count undisclosed; 1.05M-token context, reasoning effort low 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
ExploitBenchAgents99.7%#2
best: GPT-6 Astra · 100.0%
HealthBench ProfessionalKnowledge64.2%#3
best: Claude Opus 5.5 · 65.6%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$2
Output / M tok
$10
GPT-6.1 family
Elo progression across releases
API price $2/$10 · each benchmark row carries its own source badge (see methodology)