Calculator
GLM 5.3 Flash: self-host or API?
Self-hosting
Derived (an estimate)Estimate. The API is cheaper at any volume ($0.059/M at a 3:1 input:output mix vs $0.176-$1.02/M self-hosted at 50% utilization).
At 50% utilization · Cheapest API offer: $0.059 per million tokens at Pareto (3:1 input:output mix)
Hardware rental only: the hourly price the provider publishes for the machine. It does not count the engineering to deploy and run the model, monitoring, redundancy, start-up time, or storage and network costs billed separately.
- Break-even volume
- 9,018.18 million tokens a day
What published figures show
- Break-even utilization: Prudent Never (above 100%) · Median Never (above 100%) · Optimistic Never (above 100%)
- Published figure used (opens in a new tab)
Rented configuration
8 x A100 SXM 80GB at Lambda
$535.68 a day · $22.32 an hour · Guaranteed · 640 GB of GPU memory · fp8
Cheapest configuration whose weights fit: 8 x RTX PRO 6000 96GB at CoreWeave, $480.00 a day
Checked Oct 31, 2026 Price page at the provider (opens in a new tab)
Calculation detail
Throughput scenarios
| Scenario | Tokens per second | Usable capacity (million tokens a day) | Break-even utilization | Self-hosted cost per million tokens | Against the cheapest API offer |
|---|---|---|---|---|---|
| Prudent | 12,128.2 | 523.94 | Never (above 100%) | $1.02 | Dearer |
| Median | 30,320.6 | 1,309.85 | Never (above 100%) | $0.409 | Dearer |
| Optimistic | 70,564.2 | 3,048.37 | Never (above 100%) | $0.176 | Dearer |
Cheapest of 53 current standard offers, promotions, unstable prices, off-peak prices without published hours and unverified prices far below those of the other providers excluded Checked Oct 31, 2026
Assumptions
- same architecture type, scaled by memory bandwidth and active weights (x1.28)
- Selected over the cheapest configuration (RTX PRO 6000 96GB) because its throughput is derived (an estimate); it costs 11.6% more per hour, within our 25% price tolerance.
- Hardware rental only: the hourly price the provider publishes for the machine. It does not count the engineering to deploy and run the model, monitoring, redundancy, start-up time, or storage and network costs billed separately.
- Configuration: we pick the cheapest one whose weights fit. If a configuration costing up to 25% more has a measured throughput, we pick it instead; failing that, one with a derived throughput; failing that, one with a published lower bound that settles the verdict; among equals, the cheaper. A configuration whose throughput is only a wide estimate is never preferred over the cheapest. This is our rule, not a technical requirement.
- Quantization fp8; guaranteed tier prices as published by the provider; utilization 50%.
- Published figures are generally peak throughput measured at high concurrency; an interactive service with latency targets sustains less.
- Throughput is aggregate (input and output tokens, 3:1). Published figures in generated tokens are converted with a prefill/generation speed ratio of 3 (prudent), 5 (median) and 8 (optimistic). We treat a published output figure as if the GPUs only generated tokens, ignoring that the benchmark also read prompts: this understates the throughput of the whole configuration, which is the cautious side for self-hosting.
- Break-even = daily GPU cost / API price per million tokens (9,018 million tokens a day). Self-hosting is cheaper only if the average utilization you can sustain exceeds break-even / raw capacity (the break-even utilization of each scenario), whatever the number of GPUs.