Pareto holds the designated lowest glm-5.3-flash price among 49 sellers
Pareto charges $0.0375 per million Input tokens and $0.125 per million Output tokens for glm-5.3-flash, checked on November 11, 2026. Its cache read price is $0.0075. On the 3:1 mix, Pareto stands at $0.0594, and the index designates it the cheapest of 49 sellers. The gap entry for glm-5.3-flash is not under verification.
Ratio of the highest to the lowest glm-5.3-flash price across the 49 sellers compared, rounded down.
Sellers of glm-5.3-flash compared by the index in its gap entry.
The 5.8 ratio covers all 49 sellers in the entry, while the chart and the table detail the sellers the entry lists. Darkbloom follows on glm-5.3-flash at $0.065 Input and $0.18 Output, or $0.0938 on the 3:1 mix, checked on November 11, 2026. Scx-ai-gp lists $0.1285 and NanoGPT $0.15 on the 3:1 mix. Particle lists $0.175 on the same mix.
glm-5.3-flash sellers, Pareto stands alone at the low end, and several sellers cluster on identical mixed prices.
Equal mixed prices for glm-5.3-flash hide different splits and cache prices
EmpirioLabs AI, DeepInfra and Tenstorrent each list glm-5.3-flash at $0.075 Input and $0.25 Output, or $0.1188 on the 3:1 mix. Their cache read prices differ: EmpirioLabs AI charges $0.075, while DeepInfra and Tenstorrent each charge $0.015. The Tenstorrent price was checked on November 10, 2026, the other two on November 11, 2026.
Consensusprotocol and InferenceNet each reach $0.1375 on the 3:1 mix for glm-5.3-flash, with different splits. Consensusprotocol charges $0.10 Input and $0.25 Output. InferenceNet charges $0.09 Input and $0.28 Output. The sources do not say why these sellers price glm-5.3-flash differently.
| Seller | Input, $/M | Output, $/M | Mixed 3:1, $/M |
|---|---|---|---|
| Pareto | 0.0375 | 0.125 | 0.0594 |
| Darkbloom | 0.065 | 0.18 | 0.0938 |
| EmpirioLabs AI | 0.075 | 0.25 | 0.1188 |
| DeepInfra | 0.075 | 0.25 | 0.1188 |
| Tenstorrent | 0.075 | 0.25 | 0.1188 |
| Scx-ai-gp | 0.088 | 0.25 | 0.1285 |
| Consensusprotocol | 0.10 | 0.25 | 0.1375 |
| InferenceNet | 0.090 | 0.28 | 0.1375 |
| NanoGPT | 0.10 | 0.30 | 0.15 |
| Makora | 0.14 | 0.24 | 0.165 |
| Sail-research | 0.11 | 0.35 | 0.17 |
| Particle | 0.10 | 0.40 | 0.175 |
glm-5.3-flash split it differently between Input and Output.Long answers and reused prompts change which glm-5.3-flash column matters
The 3:1 mix weights Input more than Output, so a workload that generates long answers should read the Output column first. Makora lists glm-5.3-flash at $0.14 Input and $0.24 Output, checked on November 11, 2026. Particle lists $0.10 Input and $0.40 Output. Sail-research lists $0.11 Input and $0.35 Output.
A workload that sends the same long prompt many times depends on the cache read price. For glm-5.3-flash, Pareto charges $0.0075 for cache reads, Consensusprotocol $0.01 and Darkbloom $0.012. EmpirioLabs AI charges $0.075 for cache reads, its own Input price.
Moving glm-5.3-flash traffic to another seller needs a check of your own token mix
- Measure your split
Take Input, Output and cached tokens from your own billing logs for
glm-5.3-flash, and see whether the 3:1 mix describes your traffic. - Compare the right column
If your answers are long, rank sellers on Output; if your prompts repeat, rank them on cache read.
- Confirm the price
Open the seller's own pricing page: the prices here were checked on November 11, 2026, the Tenstorrent price on November 10, 2026.
- Replay a sample
Send a fixed set of your prompts to the candidate seller and to your current one; the switch passes when the answers match your quality bar.