OpenAI Answers Opus 5.5 Same-Day with GPT-6 Sol and Luna at 50% Cuts — the Price War Moves to Cost-per-Task
Within hours of Anthropic's Opus 5.5 launch on Tuesday, September 22, 2026, OpenAI released GPT-6 Sol and GPT-6 Luna — the GPT-6-family successors to the GPT-5.6 Sol/Luna tiers — cutting API prices 50% versus GPT-5.6 promotional pricing. Sol lands at $2/$10 per million tokens, "exactly half Opus 5.5's base rates"; Luna at $0.10/$0.50, "97.5% below Opus 5.5 on both raw input and output pricing" (VentureBeat). OpenAI: "Improvements in caching and inference let us serve these models at lower cost, and we're passing those savings directly on to users and customers by reducing API prices for Sol and Luna by 50 percent compared with their GPT‑5.6 promotional pricing."
The new axis: cost per completed task. Artificial Analysis found GPT-6 Sol (max) costs $1.06 per task vs $1.99 for GPT-5.6 Sol, and Luna (max) $0.07 vs $0.18. OpenAI claims "GPT‑6 Sol at xhigh effort outperforms Claude Opus 5 at max effort at just 9 percent of Opus 5's cost per task" on AutomationBench. But there is still no same-harness head-to-head: OpenAI reports Sol at 33.2% on AutomationBench 1.0.6 (xhigh, $0.27/task) while Anthropic reports an early-access Zapier eval of Opus 5.5 at 40.0% — different environments, not comparable. OpenAI also says Sol matches Fable 5.1 on FrontierCode 1.1 Main at xhigh for much less; Anthropic's own table already shows Opus 5.5 above Fable 5.1 on the same benchmark.
Context. This is the third frontier pricing move in four days — Grok 4.7's $2/$6 on Sunday (OpenAI Answers Opus 5.5 Same-Day with GPT-6 Sol and Luna at 50% Cuts — the Price War Moves to Cost-per-Task history), then the Tuesday dual launch. CNBC: the releases are "the first from either lab since fears erupted around AI's potential threat to humanity," and both labs face "stiff competition from cheaper, open-weight competitors" — naming Alibaba, Moonshot AI and DeepSeek (Xiaomi's MiMo-V2.6-Pro Takes the Open-Weights Crown — Tying Grok 4.7 on a Trillion-Parameter MoE Trained on Chinese Chips). Gizmodo's read: OpenAI is "selling them as mini-Astras," and "cost-effectiveness is becoming the most important AI benchmark." VentureBeat's closing: "The model war is no longer only about who can produce the highest score. It is increasingly about who can turn frontier intelligence into the most completed work per dollar." Sol and Luna are rolling out through ChatGPT Work and Codex; Anthropic's Opus 5.5 side of the day is in Claude Opus 5.5: The Real Story Is the Safety Architecture — 85% Fewer Containment Escapes, Silent Rerouting of Cyber Tasks.