GLM-52 897 —
GPT-56SC 873 —
CL-OP5X 865 -0.9%
GROK-46H 865 -0.9%
GEM-37FH 865 -0.9%
GPT-56T 861 —
GLM-5 856 —
MUSE-SPK 841 —
QWEN-38X 824 -2.3%
GPT-6A 820 —
KIMI-K3X 810 -1%
CL-FAB5H 787 -0.9%
CL-OP5H 764 -0.9%
CL-OP46H 742 -0.9%
CL-OP47H 733 -1.1%
GEM-38FH 676 -1%
CL-OP47 585 -0.7%
INKL 531 —
CL-OP46 496 -0.2%
CL-OP48 490 -0.2%
GLM-52 897 —
GPT-56SC 873 —
CL-OP5X 865 -0.9%
GROK-46H 865 -0.9%
GEM-37FH 865 -0.9%
GPT-56T 861 —
GLM-5 856 —
MUSE-SPK 841 —
QWEN-38X 824 -2.3%
GPT-6A 820 —
KIMI-K3X 810 -1%
CL-FAB5H 787 -0.9%
CL-OP5H 764 -0.9%
CL-OP46H 742 -0.9%
CL-OP47H 733 -1.1%
GEM-38FH 676 -1%
CL-OP47 585 -0.7%
INKL 531 —
CL-OP46 496 -0.2%
CL-OP48 490 -0.2%
← Back to feed

OpenAI Cuts GPT-5.6 Sol 20%: First Model ID Reprice Since 2022, Three-Month Window Creates API Uncertainty

OpenAI reduced GPT-5.6 Sol pricing by 20% on August 21, covering the API, Codex credits, and ChatGPT Work. The reduction applies to OpenAI’s current primary frontier API offering, and it breaks a practice the company had maintained since 2022: once a model ID is priced, it stays priced.

The discount runs for three months, through approximately November 2026. OpenAI has not committed to whether pricing reverts to prior Sol rates after that window or whether a new model replaces it before then.

What Changed and Why It Matters

OpenAI’s standard approach since GPT-4 has been to deprecate model versions and release successors at new price points, rather than repricing active model IDs mid-lifecycle. That gave API customers stable cost assumptions for production systems.

The Sol reprice breaks that convention. Developers who built cost estimation tools around Sol’s original pricing must update them. Applications that assume current pricing could face an automatic increase in November when the discount expires, depending on what OpenAI does next.

The developer forum discussion flagged a downstream problem: OpenAI has also abandoned the dated model naming concept for the 5.6 family, meaning the same model ID can carry different prices at different points in time. Documentation that archived historical pricing for audit or estimation purposes now loses its accuracy when the model ID is repriced in place.

Competitive Context

The reduction arrives as Kimi K3 open-weight and Claude Fable 5 apply the most direct competitive pressure. On LiveBench, GPT-5.6 Sol Max Effort scores 81.0 overall at $0.515 per successful task. Kimi K3 scores 79.2 at $0.348 per task. The 20% cut closes that gap materially.

The extension to Codex credits is telling. GPT-5.6 Sol holds the top slot on Terminal-Bench 2.1 at 88.8%, and keeping the cost of Codex access competitive matters to enterprise accounts evaluating AI coding agents. The price cut protects that position without requiring a new model release.

The November Question

The three-month window is the most consequential unknown. OpenAI’s developer community forum notes that by November, the company will have had three months to find “efficiencies” — implying the lower price could become permanent — or a new model generation could supersede Sol entirely, resetting the pricing conversation.

Developers with long-running API contracts should build pricing sensitivity into their cost models now rather than assume November continuity either way.