OpenRouter Cuts GPT-5.6 Sol API Pricing by 50% – Impact and Community Reaction

50% Discount on GPT-5.6 Sol Is Live

The OpenRouter marketplace now lists GPT-5.6 Sol at $2.50 per 1M input tokens and $15 per 1M output tokens, a 50 % reduction from the standard OpenAI pricing. The discount applies only to the standard OpenAI provider route on OpenRouter; other providers (Azure, Amazon Bedrock) retain their higher rates.

  • Release date: July 9 2026
  • Context window: 1 M tokens
  • Weighted average effective price: $0.889 / M input, $21.24 / M output (reflecting cache hits and discounts)
  • Uptime (last 3 days): 100 % (availability 99.80 %)
  • Best‑case throughput: 80 tokens/s (Amazon Bedrock US)
  • Best‑case latency: 2.07 s (Amazon Bedrock US)

OpenRouter’s pricing page also shows a price‑history chart indicating the new listed rates for the OpenAI provider (input $2.50, output $15) with a cache‑hit rate of 87.6 %.

Performance Highlights

Metric Best Provider Value
GPQA Diamond (graduate‑level scientific reasoning) OpenAI (auto‑routing) 91.4 %
TAU‑Bench (airline scenario) Azure (EU) 80.0 %
Tool‑call error rate Azure (US) 0.62 %
Structured‑output error rate OpenAI 1.43 %
End‑to‑end latency (average) OpenAI 8.90 s

The model ranks 96 % of OpenRouter models on the Artificial Analysis Intelligence Index, 98 % on the Coding Index, and 97 % on the Agentic Index.

Real‑World Usage Signals

Top traffic‑generating apps that route through GPT‑5.6 Sol include:

  1. Codex – a coding assistant (368 B tokens processed)
  2. Hermes Agent – an open‑source persistent AI agent (201 B tokens)
  3. pi – a personal coding agent (161 B tokens)
  4. Claude Code – Anthropic’s code‑editing tool (61.4 B tokens)
  5. OpenClaw – an AI‑driven messaging/automation agent (52.7 B tokens)

These workloads illustrate the model’s strength in multi‑step coding, tool use, and long‑context reasoning.

Community Reaction on Hacker News

The price cut generated a lively discussion (447 comments). Below are the most salient points, grouped by theme.

Praise for Capability and Cost

"Love it, its an incredibly capable model and uses far fewer tokens/time thinking. ... I may cancel my Claude subscription finally." – netsec_burn

"This is honestly impressive. Cutting the price by 50% while pushing a model this capable is exactly the kind of move that makes advanced AI feel genuinely accessible." – wahid_seddiqi

Skepticism About Value vs. Subscription Plans

"Even at these prices, switching from subsidized subscriptions to the API just isn't worth it. Not even close." – therepanic

"The title looks misleading; the discount only applies on OpenRouter, not on the native OpenAI API." – OutOfHere

Concerns Over Market Dynamics

"Price wars did wonders for many businesses, like the bike‑sharing industry in China. Overgrown datacenters or mounds of GPUs dumped into the harbour next?" – m4rtink

"This is the opening salvos of an all‑out token price war. Models are becoming a commodity; we’ll likely see prices fall toward the cost of compute." – cmiles8

"Stripe recently acquired OpenRouter for >$7 B. I’d bet that explains this move!" – dvrp

Questions About the Discount’s Scope

"Is this pricing change only for OpenRouter? I don’t see official OpenAI info about this." – krzyk

"If they cut Sol by 50 % and Luna by 80 %, the original price must have carried a massive operating margin." – z_rho_one

"My interpretation is that this is just a gimmick where they serve the OpenAI flex tier at the same discount OpenAI provides for flex and then fall back to Azure." – lyjackal

Mixed Experiences With Model Quality

"5.6‑Sol feels like a regression; it takes ages on simple problems compared to earlier models." – jeffybefffy519

"I asked it to write a todo list and got a four‑page essay; 5.4 gave a concise list." – vorpalhex

"I love the race to the bottom; if Sol isn’t the best model, it’s still up there. You don’t cut the price of the best model for no reason." – onlyrealcuzzo

Technical Clarifications

"Slightly unrelated: what's up with the 'tps' value? Does GPT‑5.6 Sol really deliver just 32 tokens/second?" – egorfine

"Are there any changes to subscription credit usage?" – matheusmoreira

What the Discount Means for Users

  1. Lower entry cost for high‑capability workloads – At $2.50 / M input tokens, developers can experiment with complex reasoning, multi‑step coding, and agentic pipelines without the $5 / M barrier that previously limited adoption.
  2. Potential shift toward OpenRouter as a primary access point – Because the discount is exclusive to OpenRouter’s routing mode, users who enable the ZDR‑only toggle (which forces OpenAI‑flex pricing) may miss out, prompting a migration of traffic to the standard route.
  3. Cache‑hit benefits – The weighted average effective input price ($0.889 / M) is far below the listed $2.50, indicating that heavy cache usage can further reduce costs for repeat prompts.
  4. Competitive pressure – Chinese models such as Kimi K3 and DeepSeek v4 are cited as comparable in quality at lower price points, suggesting OpenAI is responding to a global pricing arms race.

How to Get Started

export OPENROUTER_API_KEY=sk-or-v1-...
curl -N https://openrouter.ai/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -d '{
    "model": "openai/gpt-5.6-sol",
    "stream": true,
    "messages": [{"role": "user", "content": "Explain quantum entanglement to a 10‑year‑old"}]
  }'

The OpenRouter SDKs (TypeScript, Python, Go) support the same request format as the OpenAI API; simply swap the base URL to https://openrouter.ai/api/v1.

Bottom Line

OpenRouter’s 50 % price cut makes GPT‑5.6 Sol the most affordable flagship OpenAI model on the platform, but the discount is limited to OpenRouter’s routing tier and does not affect native OpenAI pricing. Community sentiment ranges from enthusiasm about democratizing high‑end LLM capabilities to skepticism about the long‑term sustainability of such discounts amid fierce competition and a looming token‑price war.

Sources

Related