AI pricing power is the new frontier. Grok 4.6 launched on Wednesday, matching Fable 5 Max at an 85% discount. Downloadable models set that price, a shift that moves the focus from raw capability to cost efficiency.
SpaceXAI announced the model in two sentences, calling it a “significant improvement” over Grok 4.5 at the same price. Artificial Analysis scored the new model at 61 on its Intelligence Index, five points above the previous version. On the GDPVal-AA leaderboard, Grok 4.6 hit the top at 1,753 Elo, just past Fable 5 Max at 1,741.
Related: European firms buy future AI computing power
The five-point jump came from post-training, not a larger model size. Analyst Aakash Gupta noted that Grok 4.6 runs on the same 1.5 trillion parameters as 4.5. The API pricing stayed flat at $2 in and $6 out per million tokens. Because the weights cost the same to serve, the improvement was effectively free.
When a company offers a better model for the same bill, the value is clear. The barrier to entry drops, and more organizations can afford to integrate advanced tools into their workflows.
Open Weights Force the Pricing Ceiling
Two of the three frontier launches this week included downloadable weights. This availability changes how labs can charge. When a developer has a file on Hugging Face as a fallback, a closed lab cannot simply raise prices for better performance.
Related: OpenAI says GPT-5.6 Sol reduces its own costs
Investor Gavin Baker ran the numbers and found Grok 4.6 runs 80% cheaper on input tokens and 88% cheaper on output than Fable 5 Max. He called it “Pareto dominant.” If a competitor releases a superior model at the old price, the open-source option forces a downward pressure on market rates.
Box CEO Aaron Levie explained the result. He argues that cheaper models do not shrink budgets; they fund work that was previously unaffordable. This is often called the Jevons paradox: as efficiency rises, demand rises to consume the new capacity. The value shifts to the layer that can route and optimize tasks, rather than the labs producing the models.
When models become commodities, routing becomes the business.
Related: Meta’s AI coding strategy could be reshaped by 800 errors
SpaceXAI announced Grok 4.6 in two sentences. Frontier intelligence, the company said, and a “significant improvement over Grok 4.5 at the same price.” Artificial Analysis scored it 61 on its Intelligence Index, five points above 4.5 and even with GPT-5.6 Sol. On the GDPVal-AA leaderboard it went to the top at 1,753 Elo, just past Fable 5 Max at 1,741. So where did the five points come from? Not a bigger model. As product analyst Aakash Gupta said, Grok 4.6 runs on the same 1.5 trillion parameters as 4.5 and every gain came out of post-training. Terminal-Bench jumped 66% in one release. The API price never moved, $2 in and $6 out per million tokens, because the same weights cost the same to serve. The improvement is, essentially, free.
Two of the three frontier launches this week, Aug. 11-13, 2026 Claude Fable 5Anthropic, for reference Published API pricing per 1M tokens (input / output). DeepSeek rates are off-peak; peak runs double, and the new pricing takes effect Aug. 16, 2026. Fable 5 is shown for comparison and did not ship this week. Our own Amanda Caswell went through the benchmarks SpaceXAI left out of those two sentences: 69.9% on CursorBench v3.2, a hair under Fable 5 Max at 70.5%, and behind the field on DeepSWE.
