Multiverse Computing is now the lowest-priced provider of GLM 5.3, according to Artificial Analysis. On the CompactifAI API, Z.ai's model costs $1.48 per Intelligence Index task, against $4.92 at most providers, and $0.64 per million output tokens, against $4.40.
GLM 5.3 on the CompactifAI API costs $0.02 per million cached input tokens, $0.20 per million input tokens and $0.64 per million output tokens. Output is where reasoning models spend most of their tokens, and this price is 85% below the $4.40 that most providers charge. It is also 68% below the next lowest output price in the comparison, $1.98.
Figure 1. Pricing for GLM 5.3 (max) providers, in USD per million tokens, for 10,000 input tokens. Lower is better. Source: Artificial Analysis.
On input, only Makora lists a lower price ($0.14 against $0.20), but its output price is $4.40. The cached input price of $0.02 is the lowest in the comparison, a fifth of Makora's $0.10.
What a task actually costs
Price per token only tells part of the story, because models use different numbers of tokens to finish the same work. Artificial Analysis measures the weighted average cost of completing a task in its Intelligence Index, with input, cached input, reasoning and answer tokens all included.
On that measure, GLM 5.3 from Multiverse Computing costs $1.48 per task. That is 31% below the second-lowest provider and 70% below the $4.92 where a large group of providers cluster.
Figure 2. Cost per Intelligence Index task for GLM 5.3 (max) providers, in USD, segmented by token type. Lower is better. Source: Artificial Analysis.
For teams running agents, coding assistants or document pipelines, the gap shows up directly in the monthly bill. A workload that costs $10,000 a month at $4.92 per task costs about $3,000 at $1.48.
Quasar: an optimized version of GLM 5.3, at an even lower price
Multiverse Computing also offers Quasar, an optimized version of GLM 5.3 shaped with CompactifAI, the company's optimization technology. Quasar is now available at $0.11 per million input tokens, $0.0099 per million cached input tokens and $0.33 per million output tokens. Compared with GLM 5.3 on the same API ($0.20 input, $0.018 cached input and $0.64 output), that is 45% lower on input and cached input, and 48% lower on output.
Against the rest of the GLM 5.3 market in Figure 1, the output price of $0.33 is 83% below the next lowest provider ($1.98) and 92% below the $4.40 that most providers charge.
Qwen3.8 27B: lowest price per token by a wide margin
Among the eight providers of Qwen3.8 27B in the Artificial Analysis comparison, the CompactifAI API has the lowest price on all three metrics: $0.02 per million cached tokens, $0.09 per million input tokens and $0.44 per million output tokens.
On output, the next lowest provider charges $1.88, so the Multiverse Computing price is 77% lower. Three providers charge $3.00, which makes it 85% lower. On input, the next lowest price is $0.15, so $0.09 is 40% lower, and the highest is $0.50. On cached input, the next lowest is $0.04, twice the $0.02 listed for Multiverse Computing.
Figure 3. Pricing for Qwen3.8 27B (xhigh) providers, in USD per million tokens, for 10,000 input tokens. Lower is better. Source: Artificial Analysis.
The price reduction is aligned with the mission of Multiverse Computing: making AI efficient and secure for the organizations that depend on it. Lower prices per token and per task let teams move agents, coding assistants and document pipelines from pilot to production without the bill setting the limit.
Efficiency is also the focus of Quasar and the rest of the Multiverse Computing model family, shaped with CompactifAI, the company's optimization technology.
Get started
CompactifAI API: Sign up and start building!
API documentation and pricing: docs.compactif.ai