Anthropic’s Claude Opus 5.5 and OpenAI’s GPT‑6 Sol & Luna Trigger Fresh Pricing Competition

Pricing shifts from OpenAI and Anthropic

Yesterday Anthropic launched Claude Opus 5.5 and, an hour later, OpenAI announced GPT‑6 Sol and GPT‑6 Luna.

Both companies cut prices dramatically compared with their GPT‑5.6 predecessors.

GPT‑6 Luna now costs $0.10 per million input tokens, $0.01 per million cached tokens, and $0.50 per million output tokens.

Its predecessor, GPT‑5.6 Luna, charged $0.20/$0.02/$1.20 respectively, making the new model half as expensive on every metric.

GPT‑6 Sol follows the same pattern, dropping from $4/$0.40/$20 to $2/$0.20/$10 per million tokens.

The revised pricing table shows Grok 4.7 at $2/$0.50/$6, Claude Opus 5.5 at $4/$0.20/$20, and GPT‑6 Astra at $10/$1/$50.

OpenAI has also announced a 25 % price increase for GPT‑5.6 models in November, further widening the gap.

Because GPT‑5.6 Terra was already priced like GPT‑6 Sol, the price cut makes Terra’s remaining advantages negligible.

At $0.10/$0.50, GPT‑6 Luna is among the cheapest OpenAI models ever released, rivaled only by the weaker GPT‑4.1 Nano ($0.10/$0.40) and GPT‑5 Nano ($0.05/$0.40).

Claude Opus 5.5 improvements and limitations

Claude Opus 5.5 reduces its input price to $4 per million tokens and output price to $20 per million, a 20 % cut from the $5/$25 rates of earlier Opus versions.

Cache‑read costs also fell by 60 %, a change that matters for long agentic conversations where most input tokens are cached.

Thariq Shihipar explained, “Opus 5. 5 is the result of your feedback.

It communicates clearly, it’s cheaper per token than Opus 5. 0 with the intelligence of Fable 5.

1 it’s very token efficient and works across every effort level. It’s also meant to be better at Blender.

These pricing moves align Opus 5.5’s cost with GPT‑5.6 Sol, but OpenAI’s subsequent halving of Sol’s price makes Opus 5.5 relatively more expensive.

In a novelty test, Claude Opus 5.5 at “max” thinking level failed to produce an SVG of a pelican riding a bicycle, instead looping on a self‑analysis of the request.

The model began, “This is a classic test request, so I want to plan out a well‑composed pelica…,” and never returned the expected image.

Implications for developers and the emerging price war

The new pricing puts pressure on lower‑tier models such as Anthropic’s upcoming Haiku 5.5 and Sonnet 5.5.

Haiku 4.5 currently costs $1/$5, while GPT‑6 Luna offers the same capabilities at one‑tenth the price.

Developers building cost‑sensitive applications will likely gravitate toward the cheaper Luna or Sol variants.

Anthropic’s claim that Sonnet 5.5 and Haiku 5.5 are “coming soon” suggests they intend to respond with competitive rates.

The price war also affects enterprises that rely on cached token usage, as the 60 % cache discount on Opus 5.5 can lower long‑running agent costs.

Meanwhile, Grok 4.7’s $2/$0.50/$6 rates now match GPT‑6 Sol on input pricing and approach its output cost.

These dynamics indicate a rapid shift toward cheaper, high‑performance models, prompting users to reassess which provider offers the best value for their workloads.

Stakeholders should monitor upcoming model releases and price announcements, as the competitive landscape is still evolving.

Why This Matters: The steep price cuts make advanced language models more accessible, forcing developers to choose based on cost‑efficiency rather than brand alone.