GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks and long-horizon problem solving.
Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick — Balanced (price + speed), Nitro (fastest), or Floor (cheapest).
The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.
Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).
Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.
Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.
| Source | Benchmark | Score |
|---|---|---|
| Artificial Analysis | GPT-5.6 Sol (Low) Intelligence Index | 33.5 |
| Artificial Analysis | GPT-5.6 Sol (Low) Coding Index | 69.7 |
| Artificial Analysis | GPT-5.6 Sol (Low) Agentic Index | 31.6 |
| Artificial Analysis | GPT-5.6 Sol (Low) GPQA Diamond | 89.8% |
| Artificial Analysis | GPT-5.6 Sol (Low) HLE | 39.4% |
| Artificial Analysis | GPT-5.6 Sol (Low) IFBench | 66.5% |
| Artificial Analysis | GPT-5.6 Sol (Low) τ²-Bench Telecom | 76.0% |
| Artificial Analysis | GPT-5.6 Sol (Low) AA-LCR | 78.0% |
| Artificial Analysis | GPT-5.6 Sol (Low) τ-Bench Banking | 29.1% |
| Artificial Analysis | GPT-5.6 Sol (Low) GDPval-AA | 40.1% |
| Artificial Analysis | GPT-5.6 Sol (Low) CritPt | 14.9% |
| Artificial Analysis | GPT-5.6 Sol (Low) SciCode | 56.4% |
| Artificial Analysis | GPT-5.6 Sol (Low) Terminal-Bench Hard | 60.6% |
| Artificial Analysis | GPT-5.6 Sol (Low) Terminal-Bench 4.0 | 1.0% |
| Artificial Analysis | GPT-5.6 Sol (Low) Terminal-Bench 2.1 | 76.8% |
| Artificial Analysis | GPT-5.6 Sol (Low) AA-Omniscience Accuracy | 57.2% |
| Artificial Analysis | GPT-5.6 Sol (Low) AA-Omniscience Non-Hallucination Rate | 10.6% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Intelligence Index | 44.0 |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Coding Index | 78.3 |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Agentic Index | 47.4 |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) GPQA Diamond | 93.1% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) HLE | 47.3% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) IFBench | 71.0% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) τ²-Bench Telecom | 84.8% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) AA-LCR | 82.3% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) τ-Bench Banking | 38.1% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) GDPval-AA | 53.4% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) CritPt | 28.6% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) SciCode | 57.1% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Terminal-Bench Hard | 61.4% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Terminal-Bench 4.0 | 24.7% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) Terminal-Bench 2.1 | 89.5% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) AA-Omniscience Accuracy | 58.8% |
| Artificial Analysis | GPT-5.6 Sol (Xhigh) AA-Omniscience Non-Hallucination Rate | 8.1% |
| Artificial Analysis | GPT-5.6 Sol (High) Intelligence Index | 42.3 |
| Artificial Analysis | GPT-5.6 Sol (High) Coding Index | 77.2 |
| Artificial Analysis | GPT-5.6 Sol (High) Agentic Index | 44.2 |
| Artificial Analysis | GPT-5.6 Sol (High) GPQA Diamond | 92.8% |
| Artificial Analysis | GPT-5.6 Sol (High) HLE | 46.0% |
| Artificial Analysis | GPT-5.6 Sol (High) IFBench | 69.2% |
| Artificial Analysis | GPT-5.6 Sol (High) τ²-Bench Telecom | 83.3% |
| Artificial Analysis | GPT-5.6 Sol (High) AA-LCR | 81.7% |
| Artificial Analysis | GPT-5.6 Sol (High) τ-Bench Banking | 36.7% |
| Artificial Analysis | GPT-5.6 Sol (High) GDPval-AA | 50.1% |
| Artificial Analysis | GPT-5.6 Sol (High) CritPt | 25.7% |
| Artificial Analysis | GPT-5.6 Sol (High) SciCode | 57.8% |
| Artificial Analysis | GPT-5.6 Sol (High) Terminal-Bench Hard | 62.1% |
| Artificial Analysis | GPT-5.6 Sol (High) Terminal-Bench 4.0 | 20.7% |
| Artificial Analysis | GPT-5.6 Sol (High) Terminal-Bench 2.1 | 87.3% |
| Artificial Analysis | GPT-5.6 Sol (High) AA-Omniscience Accuracy | 58.4% |
| Artificial Analysis | GPT-5.6 Sol (High) AA-Omniscience Non-Hallucination Rate | 8.8% |
| Artificial Analysis | GPT-5.6 Sol (Medium) Intelligence Index | 39.2 |
| Artificial Analysis | GPT-5.6 Sol (Medium) Coding Index | 76.3 |
| Artificial Analysis | GPT-5.6 Sol (Medium) Agentic Index | 39.7 |
| Artificial Analysis | GPT-5.6 Sol (Medium) GPQA Diamond | 92.6% |
| Artificial Analysis | GPT-5.6 Sol (Medium) HLE | 42.2% |
| Artificial Analysis | GPT-5.6 Sol (Medium) IFBench | 69.6% |
| Artificial Analysis | GPT-5.6 Sol (Medium) τ²-Bench Telecom | 81.0% |
| Artificial Analysis | GPT-5.6 Sol (Medium) AA-LCR | 80.3% |
| Artificial Analysis | GPT-5.6 Sol (Medium) τ-Bench Banking | 36.5% |
| Artificial Analysis | GPT-5.6 Sol (Medium) GDPval-AA | 46.0% |
| Artificial Analysis | GPT-5.6 Sol (Medium) CritPt | 22.9% |
| Artificial Analysis | GPT-5.6 Sol (Medium) SciCode | 57.4% |
| Artificial Analysis | GPT-5.6 Sol (Medium) Terminal-Bench Hard | 62.9% |
| Artificial Analysis | GPT-5.6 Sol (Medium) Terminal-Bench 4.0 | 14.6% |
| Artificial Analysis | GPT-5.6 Sol (Medium) Terminal-Bench 2.1 | 86.1% |
| Artificial Analysis | GPT-5.6 Sol (Medium) AA-Omniscience Accuracy | 57.8% |
| Artificial Analysis | GPT-5.6 Sol (Medium) AA-Omniscience Non-Hallucination Rate | 9.2% |
| Artificial Analysis | GPT-5.6 Sol (Max) Intelligence Index | 47.0 |
| Artificial Analysis | GPT-5.6 Sol (Max) Coding Index | 77.4 |
| Artificial Analysis | GPT-5.6 Sol (Max) Agentic Index | 50.2 |
| Artificial Analysis | GPT-5.6 Sol (Max) GPQA Diamond | 94.1% |
| Artificial Analysis | GPT-5.6 Sol (Max) HLE | 49.5% |
| Artificial Analysis | GPT-5.6 Sol (Max) IFBench | 72.7% |
| Artificial Analysis | GPT-5.6 Sol (Max) τ²-Bench Telecom | 85.1% |
| Artificial Analysis | GPT-5.6 Sol (Max) AA-LCR | 84.0% |
| Artificial Analysis | GPT-5.6 Sol (Max) τ-Bench Banking | 44.3% |
| Artificial Analysis | GPT-5.6 Sol (Max) GDPval-AA | 55.4% |
| Artificial Analysis | GPT-5.6 Sol (Max) CritPt | 32.3% |
| Artificial Analysis | GPT-5.6 Sol (Max) SciCode | 57.1% |
| Artificial Analysis | GPT-5.6 Sol (Max) Terminal-Bench Hard | 65.9% |
| Artificial Analysis | GPT-5.6 Sol (Max) Terminal-Bench 4.0 | 39.9% |
| Artificial Analysis | GPT-5.6 Sol (Max) Terminal-Bench 2.1 | 88.0% |
| Artificial Analysis | GPT-5.6 Sol (Max) AA-Omniscience Accuracy | 59.4% |
| Artificial Analysis | GPT-5.6 Sol (Max) AA-Omniscience Non-Hallucination Rate | 7.8% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) Intelligence Index | 28.3 |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) Coding Index | 65.1 |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) Agentic Index | 28.1 |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) GPQA Diamond | 79.0% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) HLE | 16.7% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) AA-LCR | 62.3% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) τ-Bench Banking | 19.6% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) GDPval-AA | 37.1% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) CritPt | 5.1% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) SciCode | 47.7% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) Terminal-Bench 2.1 | 74.2% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) AA-Omniscience Accuracy | 48.7% |
| Artificial Analysis | GPT-5.6 Sol (Non-reasoning) AA-Omniscience Non-Hallucination Rate | 7.2% |
| Vals AI | GPT-5.6 Sol Code Migration | 52.9% |
| Vals AI | GPT-5.6 Sol CorpFin v2 | 64.4% |
| Vals AI | GPT-5.6 Sol CyberBench v1.1 | 76.3% |
| Vals AI | GPT-5.6 Sol EMB | 72.3% |
| Vals AI | GPT-5.6 Sol Finance Agent (v2) | 53.8% |
| Vals AI | GPT-5.6 Sol GPQA Diamond | 95.2% |
| Vals AI | GPT-5.6 Sol Harvey's Legal Agent Benchmark v1 | 2.5% |
| Vals AI | GPT-5.6 Sol IOI | 91.2% |
| Vals AI | GPT-5.6 Sol LiveCodeBench | 82.6% |
| Vals AI | GPT-5.6 Sol LegalBench | 87.0% |
| Vals AI | GPT-5.6 Sol Legal Research Bench | 48.1% |
| Vals AI | GPT-5.6 Sol MedCode | 44.0% |
| Vals AI | GPT-5.6 Sol MedScribe | 85.2% |
| Vals AI | GPT-5.6 Sol MMLU Pro | 89.1% |
| Vals AI | GPT-5.6 Sol MMMU Pro | 88.8% |
| Vals AI | GPT-5.6 Sol MortgageTax | 67.3% |
| Vals AI | GPT-5.6 Sol ProgramBench | 1.5% |
| Vals AI | GPT-5.6 Sol ProofBench v1.1 | 83.0% |
| Vals AI | GPT-5.6 Sol Vals RSI Index | 23.9% |
| Vals AI | GPT-5.6 Sol SAGE | 52.6% |
| Vals AI | GPT-5.6 Sol Vals Science Index | 41.5% |
| Vals AI | GPT-5.6 Sol SkillsBench | 54.1% |
| Vals AI | GPT-5.6 Sol SRE Bench | 30.5% |
| Vals AI | GPT-5.6 Sol SWE-bench | 96.2% |
| Vals AI | GPT-5.6 Sol Tax Agent Bench | 31.1% |
| Vals AI | GPT-5.6 Sol TaxEval v2 | 74.8% |
| Vals AI | GPT-5.6 Sol Terminal-Bench 4.0 | 37.9% |
| Vals AI | GPT-5.6 Sol Vals Index | 58.0% |
| Vals AI | GPT-5.6 Sol Vals Multimodal Index | 72.6% |
| Vals AI | GPT-5.6 Sol Vibe Code Bench 1-100 | 20.0% |
| Vals AI | GPT-5.6 Sol Vibe Code Bench v1.1 | 80.5% |
| Vals AI | GPT-5.6 Sol Vals Web Search Index | 43.6% |
Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.
Token volume and request traffic to this model over time.
Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.
OpenAI: GPT-5.6 Sol is available through EU in-region routing. Requests sent to eu.openrouter.ai are decrypted in the EU and routed only to shared provider endpoints in Europe (the EU/EEA, the UK, and Switzerland).
https://eu.openrouter.ai/api/v1Fails closed: if no shared provider endpoint in Europe is available, the request errors instead of leaving the region. Requests served by your own BYOK key go to your deployment’s region, which you control. In-region routing docs
Guides, cookbooks, blog posts, and tools for building with OpenAI: GPT-5.6 Sol on OpenRouter.
GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks and long-horizon problem solving.
The GPT-5.6 Sol documentation on OpenRouter includes the GPT-5.6 Migration Guide cookbook and the GPT 5.6 Discounts & Jevons Paradox blog post. The quickstart covers API keys and your first request, and the API reference lists every parameter.
GPT-5.6 Sol costs $4.00/M input tokens and $20.00/M output tokens, with separate rates for Cache Read at $0.40/M tokens, Cache Write at $5.00/M tokens, and Web Search at $10.00/1K calls.
GPT-5.6 Sol has a 1,050,000 token context window. It supports up to 128,000 completion tokens.
Yes. GPT-5.6 Sol accepts tools and tool_choice for function calling. It also supports structured outputs via a JSON schema in response_format.
Yes. OpenAI caches prompts of 1024 tokens or more automatically, and GPT-5.6 Sol bills a cache write on each cached prompt whether or not the cached prefix is read again. To turn caching off for a request, set prompt_cache_options.mode to "explicit" at the request root and do not mark any content block with prompt_cache_breakpoint. Explicit mode disables OpenAI-managed breakpoints, so a request with none of your own has no cacheable prefix and is billed at the plain input rate. See the prompt caching guide.
GPT-5.6 Sol accepts files such as PDFs, images, and text as input and returns text.
GPT-5.6 Sol is served by 3 providers on OpenRouter: OpenAI, Azure, and Amazon Bedrock. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.
GPT-5.6 Sol was released on July 9, 2026. Its knowledge cutoff is February 16, 2026.