GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol, positioned between GPT-6 Astra and GPT-6 Luna, with a 1,050,000-token context window.

Pricing: Input 120 credits / 1M tokens (≈ $0.60), Cached Input 6 credits / 1M tokens (≈ $0.03), Cache Writes 150 credits / 1M tokens (≈ $0.75), Output 600 credits / 1M tokens (≈ $3.00) — all 70% below official pricing. High-tier top-ups come with bonus credits that lower the effective price by about 10%.
gpt-6-1-sol

Automatically append previous messages to maintain multi-turn context. May increase token usage.

Function calling
Web Access
Reasoning effort
Stream Response

How can I help you today?

README

Complete guide to using gpt-6-1-sol

GPT-6.1 Sol API: How It Compares with GPT-6 Astra and Luna

Original image
ParameterValue
Modelgpt-6-1-sol (responses report the model as gpt-6.1-sol)
EndpointPOST /codex/v1/responses, compatible with the OpenAI Responses API
Context window1,050,000 tokens
Maximum input922,000 tokens
Maximum output128,000 tokens
Input typesText, images and files, mixed in one message
Reasoning effortlow, medium, high, xhigh or max; default medium
ToolsWeb search or function calling, not both in one request
StreamingOn by default (server-sent events); set stream to false for a single JSON response
PricingShown in the pricing card above
GPT-6.1 SolGPT-6 AstraGPT-6 Luna
How OpenAI positions itNearly matches Astra on agentic coding, computer use and professional workIts best model across the board, for when you want the best resultsA faster, more affordable GPT-6 model trained with Astra's methods
Artificial Analysis Intelligence Index (Oct 2026)525338
LMArena text score (Oct 8, 2026)148414751443
OpenAI-reported comparison with AstraMatches Astra on DeepSWE v1.1; within 2.1 points on OSWorld 2.0 (max effort)Scored above GPT-6.1 Sol on Terminal-Bench Science 0.1 (68.1%)Not reported in the GPT-6.1 Sol release
Available on Kie.aiYesYesYes
Benchmark (OpenAI-reported)What it measuresGPT-6.1 Sol vs GPT-6 Sol
DeepSWE v1.1Complex software-engineering tasks in real codebases6.4 points above GPT-6 Sol's best score, at a lower reasoning effort
AutomationBenchMulti-step business workflows4.8 points higher at medium reasoning effort
OSWorld 2.0 (offline set)Demanding computer-use workflows7 points higher at maximum reasoning effort
Terminal-Bench Science 0.1Data analysis, simulation and theorem provingMore than double GPT-6 Sol's score at maximum reasoning effort
Factuality evaluationShare of answers with at least one factual error11.4% to 7.7% at low reasoning effort, on deliberately difficult prompts that are not representative of typical usage

Not across the board. On OpenAI's evaluations it matches Astra on some coding and computer-use work, and independent leaderboards score the two within a few points, while OpenAI still recommends Astra for the most difficult research tasks. The comparison table above has the scores.

On the Artificial Analysis Intelligence Index, GPT-6 Luna scores 38 and GPT-6.1 Sol scores 52 (as of October 2026). All three GPT-6 models are available on Kie.ai, and the comparison table above lines them up.

GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol, built for agentic coding, computer use and professional work. On Kie.ai it is the API model gpt-6-1-sol; its parameters, scores and uses are covered above.

Yes, on the results OpenAI reports: it scores higher on coding, business-workflow, computer-use and science benchmarks and makes fewer factual errors on difficult prompts. The benchmark table above lists each result.

Pricing is shown in the pricing card above.

OpenAI lists a 1,050,000-token context window with up to 128,000 output tokens. The parameter table above has the other limits.

Not on Kie.ai: each request uses credits, as shown in the pricing card above. New Kie.ai users receive 80 free credits for testing.

Low, medium, high, xhigh and max, with medium as the default.